Merge pull request 'feat(dbml): @postgres/@sqlite dialect directives (#19)' (#27) from issue-19-dbml-directives into master

Reviewed-on: #27

.gitea/workflows/release.yml

@@ -20,12 +20,34 @@ jobs:
         with:
           go-version-file: go.mod
 
+      - name: go vet
+        run: go vet ./...
+
+      - name: Install lint tools
+        run: |
+          go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@latest
+          go install honnef.co/go/tools/cmd/staticcheck@latest
+          go install golang.org/x/vuln/cmd/govulncheck@latest
+          echo "$(go env GOPATH)/bin" >> "$GITHUB_PATH"
+
+      - name: gofumpt (golangci-lint fmt)
+        run: |
+          diff=$(golangci-lint fmt --diff)
+          if [ -n "$diff" ]; then
+            echo "$diff"
+            echo "Formatting issues found. Run: make fmt"
+            exit 1
+          fi
+
+      - name: staticcheck
+        run: staticcheck ./...
+
+      - name: govulncheck
+        run: govulncheck ./...
+
       - name: Test
         run: go test ./...
 
-      - name: Lint
-        run: go vet ./...
-
   release:
     needs: test
     runs-on: ubuntu-latest
@@ -222,6 +244,7 @@ jobs:
             PKGDIR="relspec_${PKGVER}_${GOARCH}"
             mkdir -p "${PKGDIR}/DEBIAN"
             mkdir -p "${PKGDIR}/usr/bin"
+            chmod -R 0755 "${PKGDIR}"
 
             install -m755 relspec "${PKGDIR}/usr/bin/relspec"
 

.golangci.json

@@ -1,7 +1,7 @@
 {
   "formatters": {
     "enable": [
-      "gofmt",
+      "gofumpt",
       "goimports"
     ],
     "exclusions": {
@@ -13,8 +13,10 @@
       ]
     },
     "settings": {
-      "gofmt": {
-        "simplify": true
+      "gofumpt": {
+        "extra": {
+          "group-params": true
+        }
       },
       "goimports": {
         "local-prefixes": [

Makefile

@@ -1,4 +1,4 @@
-.PHONY: all build test test-unit test-integration lint coverage clean install help docker-up docker-down docker-test docker-test-integration start stop release release-version godoc
+.PHONY: all build test test-unit test-integration lint coverage clean install help docker-up docker-down docker-test docker-test-integration start stop release release-version godoc vet fmt fmt-check staticcheck govulncheck check
 
 # Binary name
 BINARY_NAME=relspec
@@ -14,6 +14,11 @@ GOGET=$(GOCMD) get
 GOMOD=$(GOCMD) mod
 GOCLEAN=$(GOCMD) clean
 
+# Tool versions (compiled on demand via `go run` so they match the local toolchain)
+GOLANGCI_LINT = go run github.com/golangci/golangci-lint/v2/cmd/golangci-lint@latest
+STATICCHECK   = go run honnef.co/go/tools/cmd/staticcheck@latest
+GOVULNCHECK   = go run golang.org/x/vuln/cmd/govulncheck@latest
+
 # Version information
 VERSION := $(shell git describe --tags --always --dirty 2>/dev/null || echo "dev")
 BUILD_DATE := $(shell date -u +"%Y-%m-%d %H:%M:%S UTC")
@@ -41,6 +46,29 @@ COMPOSE_CMD := $(shell \
 
 all: lint test build ## Run linting, tests, and build
 
+check: vet fmt-check staticcheck govulncheck ## Run vet, gofumpt check, staticcheck, and govulncheck
+
+vet: ## Run go vet
+	@echo "Running go vet..."
+	$(GOCMD) vet ./...
+
+fmt: ## Format code (gofumpt + goimports via golangci-lint)
+	@echo "Formatting..."
+	$(GOLANGCI_LINT) fmt --config=.golangci.json
+
+fmt-check: ## Check formatting (gofumpt + goimports via golangci-lint)
+	@echo "Checking formatting..."
+	@diff=$$($(GOLANGCI_LINT) fmt --diff --config=.golangci.json); \
+	if [ -n "$$diff" ]; then echo "$$diff"; echo "Run: make fmt"; exit 1; fi
+
+staticcheck: ## Run staticcheck
+	@echo "Running staticcheck..."
+	$(STATICCHECK) ./...
+
+govulncheck: ## Run govulncheck
+	@echo "Running govulncheck..."
+	$(GOVULNCHECK) ./...
+
 build: deps ## Build the binary
 	@echo "Building $(BINARY_NAME) $(VERSION)..."
 	@mkdir -p $(BUILD_DIR)

README.md

@@ -106,6 +106,49 @@ Modes: `database` (default) · `schema` · `table` · `script`
 
 Template functions: string utils (`toCamelCase`, `toSnakeCase`, `pluralize`, …), type converters (`sqlToGo`, `sqlToTypeScript`, …), filters, loop helpers, safe access.
 
+### `job` — Declarative job files
+
+Run named jobs from a `relspec.yml` manifest instead of repeating long command lines.
+
+```bash
+# List jobs discovered in ./relspec.yml and ./relspec.<name>.yml (deterministic)
+relspec job list
+
+# Validate and print the plan without running anything
+relspec job run build-schema --plan
+
+# Run a job (and its declared dependencies)
+relspec job run build-schema
+```
+
+```yaml
+# relspec.yml
+version: 1
+jobs:
+  build-schema:
+    command: convert            # closed allow-list: convert | merge | scripts-list
+    description: Merge the DBML sources and emit PostgreSQL DDL
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    output:
+      format: pgsql
+      path: build/schema.sql
+      overwrite: true
+    options:
+      flatten_schema: false
+    logfile: .relspec/log/build-schema.log
+```
+
+The job system is **not** a shell: `command` is a fixed enum, every path is
+resolved relative to the job file and may not escape it, and remote database
+credentials are referenced by environment-variable name (`conn_env:`) and
+redacted from logs. The whole plan — unknown commands/formats, duplicate job
+names, missing inputs, path traversal, dependency cycles — is validated before
+any job runs. See [docs/JOB_FILES.md](docs/JOB_FILES.md).
+
 ### `edit` — Interactive TUI editor
 
 ```bash

cmd/relspec/convert_from_list_test.go

@@ -46,7 +46,7 @@ func TestReadDatabaseListForConvert_MultipleFiles(t *testing.T) {
 
 func TestReadDatabaseListForConvert_PathWithSpaces(t *testing.T) {
 	spacedDir := filepath.Join(t.TempDir(), "my schema files")
-	if err := os.MkdirAll(spacedDir, 0755); err != nil {
+	if err := os.MkdirAll(spacedDir, 0o755); err != nil {
 		t.Fatal(err)
 	}
 	file := filepath.Join(spacedDir, "my users schema.json")
@@ -63,7 +63,7 @@ func TestReadDatabaseListForConvert_PathWithSpaces(t *testing.T) {
 
 func TestReadDatabaseListForConvert_MultipleFilesPathWithSpaces(t *testing.T) {
 	spacedDir := filepath.Join(t.TempDir(), "my schema files")
-	if err := os.MkdirAll(spacedDir, 0755); err != nil {
+	if err := os.MkdirAll(spacedDir, 0o755); err != nil {
 		t.Fatal(err)
 	}
 	file1 := filepath.Join(spacedDir, "users schema.json")
@@ -154,7 +154,7 @@ func TestRunConvert_FromListEndToEndPathWithSpaces(t *testing.T) {
 	defer restoreConvertState(saved)
 
 	spacedDir := filepath.Join(t.TempDir(), "my schema dir")
-	if err := os.MkdirAll(spacedDir, 0755); err != nil {
+	if err := os.MkdirAll(spacedDir, 0o755); err != nil {
 		t.Fatal(err)
 	}
 	file1 := filepath.Join(spacedDir, "users schema.json")

cmd/relspec/inspect.go

@@ -193,7 +193,7 @@ func runInspect(cmd *cobra.Command, args []string) error {
 
 	// Write output
 	if inspectOutputPath != "" {
-		err = os.WriteFile(inspectOutputPath, []byte(formattedReport), 0644)
+		err = os.WriteFile(inspectOutputPath, []byte(formattedReport), 0o644)
 		if err != nil {
 			return fmt.Errorf("failed to write output file: %w", err)
 		}

cmd/relspec/job.go

@@ -0,0 +1,935 @@
+package main
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"sort"
+	"strings"
+	"time"
+
+	"github.com/spf13/cobra"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/diff"
+	"git.warky.dev/wdevs/relspecgo/pkg/inspector"
+	"git.warky.dev/wdevs/relspecgo/pkg/jobs"
+	"git.warky.dev/wdevs/relspecgo/pkg/merge"
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/readers"
+	"git.warky.dev/wdevs/relspecgo/pkg/readers/sqldir"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers"
+	wpgsql "git.warky.dev/wdevs/relspecgo/pkg/writers/pgsql"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers/sqlexec"
+	wtemplate "git.warky.dev/wdevs/relspecgo/pkg/writers/template"
+)
+
+var (
+	jobDir    string
+	jobFiles  []string
+	jobDryRun bool
+	jobNoDeps bool
+)
+
+var jobCmd = &cobra.Command{
+	Use:   "job",
+	Short: "Run declarative RelSpec jobs from job files",
+	Long: `Run named jobs declared in job files instead of repeating command-line arguments.
+
+A job file is a YAML manifest (relspec.yml, or relspec.<name>.yml for extra
+files) describing one or more jobs. Each job names a RelSpec command plus its
+inputs, output and options:
+
+  version: 1
+  jobs:
+    build-schema:
+      command: convert
+      description: Merge the DBML sources and emit PostgreSQL DDL
+      inputs:
+        - path: schema/core.dbml
+          format: dbml
+        - path: schema/tenant.dbml
+          format: dbml
+      output:
+        format: pgsql
+        path: build/schema.sql
+        overwrite: true
+      options:
+        flatten_schema: false
+      logfile: .relspec/log/build-schema.log
+
+Rules and guarantees:
+  - command is a closed allow-list (convert, merge, scripts-list, templ). Arbitrary
+    shell strings are never executed.
+  - Every path is relative to the directory holding the job file and may not
+    escape it. Absolute and home-relative paths are rejected.
+  - Remote database credentials are referenced by environment-variable name
+    via conn_env; connection strings are never stored in the manifest and are
+    redacted from logs and diagnostics.
+  - Discovery and listing are deterministic.
+  - The whole plan is validated - unknown commands/formats, duplicate job
+    names, missing inputs, path traversal, dependency cycles - before any job
+    runs. Nothing is read, written or executed when validation fails.
+  - A failed job propagates the underlying non-zero exit status and writes no
+    success marker.`,
+}
+
+var jobListCmd = &cobra.Command{
+	Use:   "list",
+	Short: "List jobs discovered in job files (deterministic order)",
+	RunE:  runJobList,
+}
+
+var jobRunCmd = &cobra.Command{
+	Use:   "run <job-name>",
+	Short: "Run a named job (and its dependencies) from a job file",
+	Args:  cobra.ExactArgs(1),
+	RunE:  runJobRun,
+}
+
+func init() {
+	for _, c := range []*cobra.Command{jobListCmd, jobRunCmd} {
+		c.Flags().StringVar(&jobDir, "dir", ".", "Directory to discover job files in")
+		c.Flags().StringSliceVar(&jobFiles, "file", nil, "Explicit job file(s) to load (repeatable); disables discovery")
+	}
+	jobRunCmd.Flags().BoolVar(&jobDryRun, "dry-run", false, "Validate and print the execution plan without running anything")
+	jobRunCmd.Flags().BoolVar(&jobDryRun, "plan", false, "Alias for --dry-run")
+	jobRunCmd.Flags().BoolVar(&jobNoDeps, "no-deps", false, "Run only the named job, skipping its declared dependencies")
+
+	jobCmd.AddCommand(jobListCmd)
+	jobCmd.AddCommand(jobRunCmd)
+}
+
+// loadJobSet discovers or loads the requested job files and runs full
+// validation. The returned Set is safe to plan and execute.
+func loadJobSet() (*jobs.Set, error) {
+	paths := jobFiles
+	if len(paths) == 0 {
+		discovered, err := jobs.Discover(jobDir)
+		if err != nil {
+			return nil, err
+		}
+		paths = discovered
+	} else {
+		for i, p := range paths {
+			if _, err := os.Stat(p); err != nil {
+				return nil, fmt.Errorf("job file %q: %w", p, err)
+			}
+			paths[i] = p
+		}
+	}
+	set, err := jobs.Load(paths)
+	if err != nil {
+		return nil, err
+	}
+	if err := set.Validate(); err != nil {
+		return nil, err
+	}
+	for _, w := range set.Warnings {
+		fmt.Fprintf(os.Stderr, "warning: %s\n", w)
+	}
+	return set, nil
+}
+
+func runJobList(cmd *cobra.Command, args []string) error {
+	set, err := loadJobSet()
+	if err != nil {
+		return err
+	}
+	out := cmd.OutOrStdout()
+
+	fmt.Fprintf(os.Stderr, "\n=== RelSpec Jobs ===\n")
+	fmt.Fprintf(os.Stderr, "Job files:\n")
+	for _, f := range set.Files {
+		fmt.Fprintf(os.Stderr, "  - %s\n", f)
+	}
+	fmt.Fprintln(os.Stderr)
+
+	names := set.Names()
+	if len(names) == 0 {
+		fmt.Fprintln(out, "(no jobs defined)")
+		return nil
+	}
+
+	nameW, cmdW, srcW := len("NAME"), len("COMMAND"), len("SOURCE")
+	for _, n := range names {
+		j := set.Jobs[n]
+		nameW = maxInt(nameW, len(n))
+		cmdW = maxInt(cmdW, len(j.Command))
+		srcW = maxInt(srcW, len(j.SourceFile))
+	}
+	fmt.Fprintf(out, "%-*s  %-*s  %-*s  %s\n", nameW, "NAME", cmdW, "COMMAND", srcW, "SOURCE", "DESCRIPTION")
+	for _, n := range names {
+		j := set.Jobs[n]
+		fmt.Fprintf(out, "%-*s  %-*s  %-*s  %s\n", nameW, n, cmdW, j.Command, srcW, j.SourceFile, j.Description)
+	}
+	return nil
+}
+
+func runJobRun(cmd *cobra.Command, args []string) error {
+	set, err := loadJobSet()
+	if err != nil {
+		return err
+	}
+	return executeJobPlan(set, args[0], jobDryRun, jobNoDeps, cmd.OutOrStdout())
+}
+
+// executeJobPlan resolves the plan for name, runs pre-flight checks over
+// EVERY job in the plan, and only then executes. When dryRun is set it prints
+// the plan and returns without touching any input, output or database.
+func executeJobPlan(set *jobs.Set, name string, dryRun, noDeps bool, out io.Writer) error {
+	plan, err := set.Plan(name, !noDeps)
+	if err != nil {
+		return err
+	}
+
+	// Pre-flight: resolve and check paths, output policy and env vars for the
+	// whole plan before anything runs. A failure here means no job executes.
+	resolved := make([]*resolvedJob, len(plan))
+	byName := make(map[string]*resolvedJob, len(plan))
+	for i, j := range plan {
+		rj, perr := preflightJob(j, byName)
+		if perr != nil {
+			return fmt.Errorf("job %q: %w", j.Name, perr)
+		}
+		resolved[i] = rj
+		byName[j.Name] = rj
+	}
+
+	if dryRun {
+		fmt.Fprintf(out, "RelSpec job plan for %q (dry run - nothing executed):\n\n", name)
+		for i, rj := range resolved {
+			printResolvedJob(out, i+1, len(resolved), rj)
+		}
+		return nil
+	}
+
+	for _, rj := range resolved {
+		if err := executeResolvedJob(rj); err != nil {
+			// Propagate the underlying failure; no success marker is written.
+			return fmt.Errorf("job %q failed: %w", rj.job.Name, err)
+		}
+	}
+	fmt.Fprintf(os.Stderr, "\n=== Job %q complete ===\n", name)
+	return nil
+}
+
+// resolvedJob is a job with every manifest path turned into a checked
+// absolute filesystem path and every conn_env resolved to its value.
+type resolvedJob struct {
+	job           *jobs.Job
+	root          string
+	inputs        []resolvedInput
+	scriptDirs    []string
+	outputPath    string // "" when the output is a database
+	outputConn    string // resolved connection string (secret)
+	outputConnEnv string
+	logPath       string
+	logPolicy     jobs.LogPolicy
+	templatePath  string
+	reportPath    string // "" for a diff summary written to the log
+	reportFormat  string
+	rulesPath     string // "" means inspector defaults
+	selection     *splitSelection
+	secrets       []string // resolved secret values to redact from logs
+}
+
+type resolvedInput struct {
+	format  string
+	path    string // "" when the input is a database
+	conn    string // resolved connection string (secret)
+	connEnv string
+	fromJob string // producer job name when this input came from from_job
+}
+
+func preflightJob(j *jobs.Job, resolvedByName map[string]*resolvedJob) (*resolvedJob, error) {
+	root := j.Dir()
+	rj := &resolvedJob{job: j, root: root, logPolicy: j.ResolvedLogPolicy()}
+
+	if j.Logfile != "" {
+		p, err := jobs.SafeJoin(root, j.Logfile)
+		if err != nil {
+			return nil, fmt.Errorf("logfile: %w", err)
+		}
+		rj.logPath = p
+	}
+	if j.Template != "" {
+		p, err := jobs.SafeJoin(root, j.Template)
+		if err != nil {
+			return nil, fmt.Errorf("template: %w", err)
+		}
+		info, err := os.Stat(p)
+		if err != nil || info.IsDir() {
+			return nil, fmt.Errorf("template %q: not found or is a directory", j.Template)
+		}
+		rj.templatePath = p
+	}
+
+	for i, in := range j.Inputs {
+		ri := resolvedInput{format: strings.ToLower(in.Format)}
+		if in.FromJob != "" {
+			producer, ok := resolvedByName[in.FromJob]
+			if !ok {
+				return nil, fmt.Errorf("input[%d]: from_job %q is not in this plan (do not use --no-deps with from_job inputs)", i, in.FromJob)
+			}
+			if producer.outputPath == "" {
+				return nil, fmt.Errorf("input[%d]: from_job %q does not write a file output", i, in.FromJob)
+			}
+			ri.path = producer.outputPath
+			ri.format = strings.ToLower(producer.job.Output.Format)
+			ri.fromJob = in.FromJob
+			rj.inputs = append(rj.inputs, ri)
+			continue
+		}
+		if in.ConnEnv != "" {
+			v, ok := os.LookupEnv(in.ConnEnv)
+			if !ok || v == "" {
+				return nil, fmt.Errorf("input[%d]: environment variable %q (conn_env) is not set", i, in.ConnEnv)
+			}
+			ri.conn = v
+			ri.connEnv = in.ConnEnv
+			rj.secrets = append(rj.secrets, v)
+		} else {
+			p, err := jobs.SafeJoin(root, in.Path)
+			if err != nil {
+				return nil, fmt.Errorf("input[%d]: %w", i, err)
+			}
+			info, err := os.Stat(p)
+			if err != nil {
+				return nil, fmt.Errorf("input[%d]: %s: file not found", i, in.Path)
+			}
+			if info.IsDir() {
+				return nil, fmt.Errorf("input[%d]: %s: is a directory, not a file", i, in.Path)
+			}
+			ri.path = p
+		}
+		rj.inputs = append(rj.inputs, ri)
+	}
+
+	for _, d := range j.ScriptDirs {
+		p, err := jobs.SafeJoin(root, d)
+		if err != nil {
+			return nil, fmt.Errorf("script_dir %q: %w", d, err)
+		}
+		info, err := os.Stat(p)
+		if err != nil {
+			return nil, fmt.Errorf("script_dir %q: not found", d)
+		}
+		if !info.IsDir() {
+			return nil, fmt.Errorf("script_dir %q: not a directory", d)
+		}
+		rj.scriptDirs = append(rj.scriptDirs, p)
+	}
+
+	if j.Output != nil {
+		if j.Output.ConnEnv != "" {
+			v, ok := os.LookupEnv(j.Output.ConnEnv)
+			if !ok || v == "" {
+				return nil, fmt.Errorf("output: environment variable %q (conn_env) is not set", j.Output.ConnEnv)
+			}
+			rj.outputConn = v
+			rj.outputConnEnv = j.Output.ConnEnv
+			rj.secrets = append(rj.secrets, v)
+		} else {
+			p, err := jobs.SafeJoin(root, j.Output.Path)
+			if err != nil {
+				return nil, fmt.Errorf("output: %w", err)
+			}
+			if _, err := os.Stat(p); err == nil && !j.Output.Overwrite {
+				return nil, fmt.Errorf("output %s already exists (set output.overwrite: true to replace it)", j.Output.Path)
+			}
+			rj.outputPath = p
+		}
+	}
+
+	if j.Rules != "" {
+		p, err := jobs.SafeJoin(root, j.Rules)
+		if err != nil {
+			return nil, fmt.Errorf("rules: %w", err)
+		}
+		info, err := os.Stat(p)
+		if err != nil || info.IsDir() {
+			return nil, fmt.Errorf("rules %q: not found or is a directory", j.Rules)
+		}
+		rj.rulesPath = p
+	}
+
+	if j.Report != nil {
+		rj.reportFormat = strings.ToLower(j.Report.Format)
+		if j.Report.Path != "" {
+			p, err := jobs.SafeJoin(root, j.Report.Path)
+			if err != nil {
+				return nil, fmt.Errorf("report: %w", err)
+			}
+			if _, err := os.Stat(p); err == nil && !j.Report.Overwrite {
+				return nil, fmt.Errorf("report %s already exists (set report.overwrite: true to replace it)", j.Report.Path)
+			}
+			rj.reportPath = p
+		}
+	}
+
+	if j.Select != nil {
+		rj.selection = &splitSelection{
+			Schemas:        j.Select.Schemas,
+			Tables:         j.Select.Tables,
+			ExcludeSchemas: j.Select.ExcludeSchemas,
+			ExcludeTables:  j.Select.ExcludeTables,
+			DatabaseName:   j.Select.DatabaseName,
+		}
+	}
+
+	return rj, nil
+}
+
+func printResolvedJob(out io.Writer, n, total int, rj *resolvedJob) {
+	j := rj.job
+	fmt.Fprintf(out, "[%d/%d] %s\n", n, total, j.Name)
+	fmt.Fprintf(out, "      command: %s\n", j.Command)
+	if j.Description != "" {
+		fmt.Fprintf(out, "      description: %s\n", j.Description)
+	}
+	fmt.Fprintf(out, "      job file: %s\n", j.SourceFile)
+	for _, ri := range rj.inputs {
+		switch {
+		case ri.fromJob != "":
+			fmt.Fprintf(out, "      input: %s (%s) from job %q\n", ri.path, ri.format, ri.fromJob)
+		case ri.path != "":
+			fmt.Fprintf(out, "      input: %s (%s)\n", ri.path, ri.format)
+		default:
+			fmt.Fprintf(out, "      input: env:%s (%s)\n", ri.connEnv, ri.format)
+		}
+	}
+	for _, d := range rj.scriptDirs {
+		fmt.Fprintf(out, "      script dir: %s\n", d)
+	}
+	if rj.outputPath != "" {
+		fmt.Fprintf(out, "      output: %s (%s)\n", rj.outputPath, j.Output.Format)
+	} else if rj.outputConnEnv != "" {
+		fmt.Fprintf(out, "      output: env:%s (%s)\n", rj.outputConnEnv, j.Output.Format)
+	}
+	if j.Report != nil {
+		format := valueOr(rj.reportFormat, "default")
+		if rj.reportPath != "" {
+			fmt.Fprintf(out, "      report: %s (%s)\n", rj.reportPath, format)
+		} else {
+			fmt.Fprintf(out, "      report: (log) (%s)\n", format)
+		}
+	}
+	if rj.rulesPath != "" {
+		fmt.Fprintf(out, "      rules: %s\n", rj.rulesPath)
+	} else if j.Command == jobs.CommandInspect {
+		fmt.Fprintf(out, "      rules: (built-in defaults)\n")
+	}
+	if rj.selection != nil {
+		fmt.Fprintf(out, "      select: %s\n", rj.selection.summary())
+	}
+	if rj.logPath != "" {
+		fmt.Fprintf(out, "      logfile: %s (rotate >= %d bytes, keep %d)\n", rj.logPath, rj.logPolicy.MaxSizeBytes, rj.logPolicy.Keep)
+	}
+	fmt.Fprintln(out)
+}
+
+// executeResolvedJob runs a single already-validated job.
+func executeResolvedJob(rj *resolvedJob) (err error) {
+	lg, closeLog, lerr := newJobLogger(rj.logPath, rj.logPolicy, rj.secrets)
+	if lerr != nil {
+		return lerr
+	}
+	defer func() { closeLog(err) }()
+
+	lg.logf("=== job %q (%s) started at %s ===", rj.job.Name, rj.job.Command, time.Now().Format(time.RFC3339))
+
+	switch rj.job.Command {
+	case jobs.CommandConvert:
+		err = runConvertJob(rj, lg)
+	case jobs.CommandMerge:
+		err = runMergeJob(rj, lg)
+	case jobs.CommandScriptsList:
+		err = runScriptsListJob(rj, lg)
+	case jobs.CommandTempl:
+		err = runTemplJob(rj, lg)
+	case jobs.CommandSplit:
+		err = runSplitJob(rj, lg)
+	case jobs.CommandInspect:
+		err = runInspectJob(rj, lg)
+	case jobs.CommandDiff:
+		err = runDiffJob(rj, lg)
+	case jobs.CommandScriptsExec:
+		err = runScriptsExecJob(rj, lg)
+	default:
+		err = fmt.Errorf("unsupported command %q", rj.job.Command)
+	}
+	if err != nil {
+		lg.logf("FAILED: %v", err)
+	} else {
+		lg.logf("OK")
+	}
+	return err
+}
+
+func runTemplJob(rj *resolvedJob, lg *jobLogger) error {
+	db, err := readJobInputs(rj, lg)
+	if err != nil {
+		return err
+	}
+	if schema := rj.job.Options.Schema; schema != "" {
+		found := false
+		for _, s := range db.Schemas {
+			if s.Name == schema {
+				db.Schemas = []*models.Schema{s}
+				found = true
+				break
+			}
+		}
+		if !found {
+			return fmt.Errorf("schema not found: %s", schema)
+		}
+	}
+	mode := rj.job.Mode
+	if mode == "" {
+		mode = "database"
+	}
+	pattern := rj.job.FilenamePattern
+	if pattern == "" {
+		pattern = "{{.Name}}.txt"
+	}
+	writer, err := wtemplate.NewWriter(&writers.WriterOptions{
+		OutputPath: rj.outputPath,
+		Metadata: map[string]interface{}{
+			"template_path":    rj.templatePath,
+			"mode":             mode,
+			"filename_pattern": pattern,
+		},
+	})
+	if err != nil {
+		return fmt.Errorf("create template writer: %w", err)
+	}
+	lg.logf("applying template: %s (mode %s)", rj.templatePath, mode)
+	if err := writer.WriteDatabase(db); err != nil {
+		return fmt.Errorf("execute template: %w", err)
+	}
+	return nil
+}
+
+func runConvertJob(rj *resolvedJob, lg *jobLogger) error {
+	db, err := readJobInputs(rj, lg)
+	if err != nil {
+		return err
+	}
+	return writeJobOutput(rj, db, lg)
+}
+
+func runMergeJob(rj *resolvedJob, lg *jobLogger) error {
+	opts := &merge.MergeOptions{
+		SkipDomains:   rj.job.Options.SkipDomains,
+		SkipRelations: rj.job.Options.SkipRelations,
+		SkipEnums:     rj.job.Options.SkipEnums,
+		SkipViews:     rj.job.Options.SkipViews,
+		SkipSequences: rj.job.Options.SkipSequences,
+	}
+	var base *models.Database
+	for i, ri := range rj.inputs {
+		db, err := readOneJobInput(ri)
+		if err != nil {
+			return fmt.Errorf("input[%d]: %w", i, err)
+		}
+		if base == nil {
+			base = db
+			lg.logf("merge target: %s", inputLabel(ri))
+			continue
+		}
+		lg.logf("merging: %s", inputLabel(ri))
+		merge.MergeDatabases(base, db, opts)
+	}
+	base.UpdateDate()
+	return writeJobOutput(rj, base, lg)
+}
+
+func runScriptsListJob(rj *resolvedJob, lg *jobLogger) error {
+	type row struct {
+		priority int
+		sequence uint
+		name     string
+		dir      string
+		lines    int
+	}
+	var rows []row
+	for _, dir := range rj.scriptDirs {
+		reader := sqldir.NewReader(&readers.ReaderOptions{
+			FilePath: dir,
+			Metadata: map[string]any{
+				"schema_name":   valueOr(rj.job.Options.Schema, "public"),
+				"database_name": "database",
+			},
+		})
+		db, err := reader.ReadDatabase()
+		if err != nil {
+			return fmt.Errorf("%s: %w", dir, err)
+		}
+		if len(db.Schemas) == 0 {
+			continue
+		}
+		for _, s := range db.Schemas[0].Scripts {
+			lines := strings.Count(s.SQL, "\n")
+			if len(s.SQL) > 0 && !strings.HasSuffix(s.SQL, "\n") {
+				lines++
+			}
+			rows = append(rows, row{s.Priority, s.Sequence, s.Name, dir, lines})
+		}
+	}
+	sort.Slice(rows, func(i, j int) bool {
+		if rows[i].priority != rows[j].priority {
+			return rows[i].priority < rows[j].priority
+		}
+		if rows[i].sequence != rows[j].sequence {
+			return rows[i].sequence < rows[j].sequence
+		}
+		if rows[i].name != rows[j].name {
+			return rows[i].name < rows[j].name
+		}
+		return rows[i].dir < rows[j].dir
+	})
+	lg.logf("found %d script(s) across %d director(y/ies):", len(rows), len(rj.scriptDirs))
+	lg.logf("%-4s %-9s %-9s %-30s %-6s %s", "No.", "Priority", "Sequence", "Name", "Lines", "Directory")
+	for i, r := range rows {
+		lg.logf("%-4d %-9d %-9d %-30s %-6d %s", i+1, r.priority, r.sequence, r.name, r.lines, r.dir)
+	}
+	return nil
+}
+
+func runSplitJob(rj *resolvedJob, lg *jobLogger) error {
+	db, err := readJobInputs(rj, lg)
+	if err != nil {
+		return err
+	}
+	sel := splitSelection{}
+	if rj.selection != nil {
+		sel = *rj.selection
+	}
+	filtered, err := filterDatabaseSelection(db, sel)
+	if err != nil {
+		return fmt.Errorf("split selection: %w", err)
+	}
+	if sel.DatabaseName != "" {
+		filtered.Name = sel.DatabaseName
+	}
+	tables := 0
+	for _, s := range filtered.Schemas {
+		tables += len(s.Tables)
+	}
+	lg.logf("split: selected %d schema(s), %d table(s)", len(filtered.Schemas), tables)
+	return writeJobOutput(rj, filtered, lg)
+}
+
+func runInspectJob(rj *resolvedJob, lg *jobLogger) error {
+	db, err := readJobInputs(rj, lg)
+	if err != nil {
+		return err
+	}
+	config, err := inspector.LoadConfig(rj.rulesPath) // "" -> built-in defaults
+	if err != nil {
+		return fmt.Errorf("load rules: %w", err)
+	}
+	report, err := inspector.NewInspector(db, config).Inspect()
+	if err != nil {
+		return fmt.Errorf("inspection failed: %w", err)
+	}
+
+	var formatted string
+	switch valueOr(rj.reportFormat, "markdown") {
+	case "json":
+		formatted, err = inspector.NewJSONFormatter().Format(report)
+	default:
+		formatted, err = inspector.NewMarkdownFormatter(io.Discard).Format(report)
+	}
+	if err != nil {
+		return fmt.Errorf("format report: %w", err)
+	}
+	if werr := atomicWrite(rj.reportPath, func(tmp string) error {
+		return os.WriteFile(tmp, []byte(formatted), 0o644)
+	}); werr != nil {
+		return werr
+	}
+	lg.logf("inspect: %d error(s), %d warning(s) -> %s",
+		report.Summary.ErrorCount, report.Summary.WarningCount, rj.reportPath)
+	if report.HasErrors() {
+		return fmt.Errorf("inspection found %d error(s)", report.Summary.ErrorCount)
+	}
+	return nil
+}
+
+func runDiffJob(rj *resolvedJob, lg *jobLogger) error {
+	if len(rj.inputs) != 2 {
+		return fmt.Errorf("diff requires exactly 2 inputs, got %d", len(rj.inputs))
+	}
+	source, err := readOneJobInput(rj.inputs[0])
+	if err != nil {
+		return fmt.Errorf("input[0]: %w", err)
+	}
+	lg.logf("diff source: %s", inputLabel(rj.inputs[0]))
+	target, err := readOneJobInput(rj.inputs[1])
+	if err != nil {
+		return fmt.Errorf("input[1]: %w", err)
+	}
+	lg.logf("diff target: %s", inputLabel(rj.inputs[1]))
+
+	result := diff.CompareDatabases(source, target)
+	s := diff.ComputeSummary(result)
+	lg.logf("diff: schemas %d/%d/%d, tables %d/%d/%d, columns %d/%d/%d (missing/extra/modified)",
+		s.Schemas.Missing, s.Schemas.Extra, s.Schemas.Modified,
+		s.Tables.Missing, s.Tables.Extra, s.Tables.Modified,
+		s.Columns.Missing, s.Columns.Extra, s.Columns.Modified)
+
+	format := diff.FormatSummary
+	switch rj.reportFormat {
+	case "json":
+		format = diff.FormatJSON
+	case "html":
+		format = diff.FormatHTML
+	}
+
+	if rj.reportPath == "" {
+		var buf bytes.Buffer
+		if err := diff.FormatDiff(result, format, &buf); err != nil {
+			return fmt.Errorf("format diff: %w", err)
+		}
+		for _, line := range strings.Split(strings.TrimRight(buf.String(), "\n"), "\n") {
+			lg.logf("%s", line)
+		}
+		return nil
+	}
+	if werr := atomicWrite(rj.reportPath, func(tmp string) error {
+		f, err := os.Create(tmp)
+		if err != nil {
+			return err
+		}
+		defer f.Close()
+		return diff.FormatDiff(result, format, f)
+	}); werr != nil {
+		return werr
+	}
+	lg.logf("diff report written: %s", rj.reportPath)
+	return nil
+}
+
+func runScriptsExecJob(rj *resolvedJob, lg *jobLogger) error {
+	schemaName := valueOr(rj.job.Options.Schema, "public")
+	combined := &models.Schema{Name: schemaName}
+	for _, dir := range rj.scriptDirs {
+		reader := sqldir.NewReader(&readers.ReaderOptions{
+			FilePath: dir,
+			Metadata: map[string]any{
+				"schema_name":   schemaName,
+				"database_name": "database",
+			},
+		})
+		db, err := reader.ReadDatabase()
+		if err != nil {
+			return fmt.Errorf("%s: %w", dir, err)
+		}
+		if len(db.Schemas) == 0 {
+			continue
+		}
+		combined.Scripts = append(combined.Scripts, db.Schemas[0].Scripts...)
+	}
+	if len(combined.Scripts) == 0 {
+		lg.logf("no scripts found; nothing to execute")
+		return nil
+	}
+	lg.logf("executing %d script(s) against database env:%s", len(combined.Scripts), rj.outputConnEnv)
+
+	writer := sqlexec.NewWriter(&writers.WriterOptions{
+		Metadata: map[string]any{
+			"connection_string": rj.outputConn,
+			"ignore_errors":     rj.job.Options.ContinueOnError,
+		},
+	})
+	if err := writer.WriteSchema(combined); err != nil {
+		return fmt.Errorf("script execution failed: %w", err)
+	}
+
+	opts := writer.Options()
+	total, _ := opts.Metadata["execution_total"].(int)
+	success, _ := opts.Metadata["execution_success"].(int)
+	failed, _ := opts.Metadata["execution_failed"].(int)
+	lg.logf("executed %d script(s): %d succeeded, %d failed", total, success, failed)
+	if failed > 0 && !rj.job.Options.ContinueOnError {
+		return fmt.Errorf("%d script(s) failed", failed)
+	}
+	return nil
+}
+
+// readJobInputs reads every input and additively merges them into one model.
+func readJobInputs(rj *resolvedJob, lg *jobLogger) (*models.Database, error) {
+	var base *models.Database
+	for i, ri := range rj.inputs {
+		db, err := readOneJobInput(ri)
+		if err != nil {
+			return nil, fmt.Errorf("input[%d]: %w", i, err)
+		}
+		lg.logf("read input: %s", inputLabel(ri))
+		if base == nil {
+			base = db
+		} else {
+			merge.MergeDatabases(base, db, &merge.MergeOptions{})
+		}
+	}
+	if base == nil {
+		return nil, fmt.Errorf("no inputs produced a database")
+	}
+	return base, nil
+}
+
+func readOneJobInput(ri resolvedInput) (*models.Database, error) {
+	if ri.conn != "" {
+		return readDatabaseForConvert(ri.format, "", ri.conn)
+	}
+	return readDatabaseForConvert(ri.format, ri.path, "")
+}
+
+func inputLabel(ri resolvedInput) string {
+	if ri.path != "" {
+		return fmt.Sprintf("%s (%s)", ri.path, ri.format)
+	}
+	return fmt.Sprintf("env:%s (%s)", ri.connEnv, ri.format)
+}
+
+// writeJobOutput writes db to the job's output target (file or database).
+func writeJobOutput(rj *resolvedJob, db *models.Database, lg *jobLogger) error {
+	o := rj.job.Options
+	format := strings.ToLower(rj.job.Output.Format)
+
+	if rj.outputConn != "" {
+		if format != "pgsql" {
+			return fmt.Errorf("database output is only supported for pgsql (got %q)", rj.job.Output.Format)
+		}
+		lg.logf("writing output to database env:%s", rj.outputConnEnv)
+		writerOpts := newWriterOptions("", o.Package, o.FlattenSchema, "", "", o.ContinueOnError)
+		writerOpts.Metadata = map[string]interface{}{"connection_string": rj.outputConn}
+		return wpgsql.NewWriter(writerOpts).WriteDatabase(db)
+	}
+
+	if err := os.MkdirAll(filepath.Dir(rj.outputPath), 0o755); err != nil {
+		return fmt.Errorf("failed to create output directory: %w", err)
+	}
+	lg.logf("writing output: %s (%s)", rj.outputPath, format)
+
+	write := func(target string) error {
+		return writeDatabase(db, format, target, o.Package, o.Schema, o.FlattenSchema, "", "", o.ContinueOnError, "")
+	}
+	// Single-file formats are written to a temp file and renamed into place so
+	// a failure never leaves a partial or truncated output. Directory-emitting
+	// formats (gorm/bun/drizzle/typeorm/prisma) write in place.
+	if jobs.SingleFileOutputFormat(format) {
+		return atomicWrite(rj.outputPath, write)
+	}
+	return write(rj.outputPath)
+}
+
+// atomicWrite calls produce with a temp path in the same directory as
+// finalPath, then renames it over finalPath. The temp file is removed on any
+// error so the destination is only ever replaced by a complete file.
+func atomicWrite(finalPath string, produce func(tmpPath string) error) error {
+	dir := filepath.Dir(finalPath)
+	if err := os.MkdirAll(dir, 0o755); err != nil {
+		return fmt.Errorf("failed to create output directory: %w", err)
+	}
+	tmp := filepath.Join(dir, fmt.Sprintf(".%s.relspec-tmp-%d", filepath.Base(finalPath), os.Getpid()))
+	if err := produce(tmp); err != nil {
+		_ = os.Remove(tmp)
+		return err
+	}
+	if err := os.Rename(tmp, finalPath); err != nil {
+		_ = os.Remove(tmp)
+		return fmt.Errorf("failed to finalize %s: %w", finalPath, err)
+	}
+	return nil
+}
+
+// --- logging + redaction ---------------------------------------------------
+
+type jobLogger struct {
+	file    io.Writer
+	secrets []string
+}
+
+// newJobLogger returns a logger that mirrors to stderr and, when path is set,
+// to a job logfile. Connection strings and known secret values are redacted
+// from everything it writes.
+func newJobLogger(path string, policy jobs.LogPolicy, secrets []string) (*jobLogger, func(err error), error) {
+	lg := &jobLogger{secrets: secrets}
+	if path == "" {
+		return lg, func(error) {}, nil
+	}
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		return nil, nil, fmt.Errorf("failed to create log directory: %w", err)
+	}
+	rotateLogIfNeeded(path, policy)
+	f, err := os.OpenFile(path, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0o644)
+	if err != nil {
+		return nil, nil, fmt.Errorf("failed to open logfile %q: %w", path, err)
+	}
+	lg.file = f
+	return lg, func(runErr error) {
+		if runErr != nil {
+			fmt.Fprintf(f, "%s job ended with error\n", time.Now().Format(time.RFC3339))
+		}
+		_ = f.Close()
+	}, nil
+}
+
+// rotateLogIfNeeded renames path -> path.1 -> path.2 ... up to policy.Keep
+// when path has grown to policy.MaxSizeBytes or more. The oldest file beyond
+// Keep is deleted. A zero/negative MaxSizeBytes disables rotation.
+func rotateLogIfNeeded(path string, policy jobs.LogPolicy) {
+	if policy.MaxSizeBytes <= 0 {
+		return
+	}
+	info, err := os.Stat(path)
+	if err != nil || info.Size() < policy.MaxSizeBytes {
+		return
+	}
+	if policy.Keep < 1 {
+		_ = os.Remove(path)
+		return
+	}
+	_ = os.Remove(fmt.Sprintf("%s.%d", path, policy.Keep))
+	for i := policy.Keep - 1; i >= 1; i-- {
+		_ = os.Rename(fmt.Sprintf("%s.%d", path, i), fmt.Sprintf("%s.%d", path, i+1))
+	}
+	_ = os.Rename(path, path+".1")
+}
+
+func (l *jobLogger) logf(format string, args ...interface{}) {
+	line := l.redact(fmt.Sprintf(format, args...))
+	fmt.Fprintf(os.Stderr, "      %s\n", line)
+	if l.file != nil {
+		fmt.Fprintf(l.file, "%s %s\n", time.Now().Format(time.RFC3339), line)
+	}
+}
+
+func (l *jobLogger) redact(s string) string {
+	for _, sec := range l.secrets {
+		if sec != "" {
+			s = strings.ReplaceAll(s, sec, "***")
+		}
+	}
+	return maskPassword(s)
+}
+
+// --- small helpers -------------------------------------------------------
+
+func maxInt(a, b int) int {
+	if a > b {
+		return a
+	}
+	return b
+}
+
+func valueOr(v, def string) string {
+	if v == "" {
+		return def
+	}
+	return v
+}

cmd/relspec/job_test.go

@@ -0,0 +1,684 @@
+package main
+
+import (
+	"bytes"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/spf13/cobra"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/jobs"
+)
+
+func writeFile(t *testing.T, path, content string) {
+	t.Helper()
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(path, []byte(content), 0o644); err != nil {
+		t.Fatal(err)
+	}
+}
+
+// jobFixture creates a job-file project with two DBML sources and returns the
+// project directory.
+func jobFixture(t *testing.T, manifest string) string {
+	t.Helper()
+	dir := t.TempDir()
+	writeFile(t, filepath.Join(dir, "schema", "core.dbml"), "Table users {\n  id int [pk]\n  name varchar\n}\n")
+	writeFile(t, filepath.Join(dir, "schema", "tenant.dbml"), "Table posts {\n  id int [pk]\n  title varchar\n}\n")
+	writeFile(t, filepath.Join(dir, "relspec.yml"), manifest)
+	return dir
+}
+
+func mustLoadSet(t *testing.T, files ...string) *jobs.Set {
+	t.Helper()
+	set, err := jobs.Load(files)
+	if err != nil {
+		t.Fatalf("load: %v", err)
+	}
+	if err := set.Validate(); err != nil {
+		t.Fatalf("validate: %v", err)
+	}
+	return set
+}
+
+const convertMergeManifest = `version: 1
+jobs:
+  build-schema:
+    command: convert
+    description: Merge DBML sources to PostgreSQL DDL
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    output:
+      format: pgsql
+      path: build/schema.sql
+      overwrite: true
+    logfile: .relspec/log/build.log
+`
+
+func TestJobRun_ConvertMultiFileMerge(t *testing.T) {
+	dir := jobFixture(t, convertMergeManifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	if err := executeJobPlan(set, "build-schema", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("executeJobPlan: %v", err)
+	}
+
+	out, err := os.ReadFile(filepath.Join(dir, "build", "schema.sql"))
+	if err != nil {
+		t.Fatalf("expected output file: %v", err)
+	}
+	sql := string(out)
+	if !strings.Contains(sql, "users") || !strings.Contains(sql, "posts") {
+		t.Fatalf("merged output missing tables:\n%s", sql)
+	}
+
+	logData, err := os.ReadFile(filepath.Join(dir, ".relspec", "log", "build.log"))
+	if err != nil {
+		t.Fatalf("expected logfile: %v", err)
+	}
+	if !strings.Contains(string(logData), "OK") {
+		t.Fatalf("logfile missing success marker:\n%s", logData)
+	}
+}
+
+func TestJobRun_DryRunDoesNotExecute(t *testing.T) {
+	dir := jobFixture(t, convertMergeManifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	var buf bytes.Buffer
+	if err := executeJobPlan(set, "build-schema", true, false, &buf); err != nil {
+		t.Fatalf("dry run error: %v", err)
+	}
+	if !strings.Contains(buf.String(), "dry run") {
+		t.Fatalf("expected dry-run banner, got: %s", buf.String())
+	}
+	if _, err := os.Stat(filepath.Join(dir, "build", "schema.sql")); !os.IsNotExist(err) {
+		t.Fatal("dry run must not create the output file")
+	}
+	if _, err := os.Stat(filepath.Join(dir, ".relspec", "log", "build.log")); !os.IsNotExist(err) {
+		t.Fatal("dry run must not create the logfile")
+	}
+}
+
+func TestJobRun_ValidationFailureNoExecution(t *testing.T) {
+	badManifest := `version: 1
+jobs:
+  evil:
+    command: convert
+    inputs:
+      - path: ../../../etc/passwd
+        format: dbml
+    output:
+      format: json
+      path: build/out.json
+    logfile: .relspec/evil.log
+`
+	dir := jobFixture(t, badManifest)
+	if _, err := jobs.Load([]string{filepath.Join(dir, "relspec.yml")}); err != nil {
+		// structural load ok; validation should reject
+		t.Fatalf("unexpected load error: %v", err)
+	}
+	set, _ := jobs.Load([]string{filepath.Join(dir, "relspec.yml")})
+	if err := set.Validate(); err == nil {
+		t.Fatal("expected validation failure for path traversal")
+	}
+	// Nothing should have been produced.
+	if _, err := os.Stat(filepath.Join(dir, "build")); !os.IsNotExist(err) {
+		t.Fatal("validation failure must not create output dir")
+	}
+	if _, err := os.Stat(filepath.Join(dir, ".relspec")); !os.IsNotExist(err) {
+		t.Fatal("validation failure must not create logfile dir")
+	}
+}
+
+func TestJobRun_MissingInputNoExecution(t *testing.T) {
+	manifest := `version: 1
+jobs:
+  x:
+    command: convert
+    inputs:
+      - path: schema/does-not-exist.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/out.json
+    logfile: .relspec/x.log
+`
+	dir := jobFixture(t, manifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	err := executeJobPlan(set, "x", false, false, &bytes.Buffer{})
+	if err == nil || !strings.Contains(err.Error(), "not found") {
+		t.Fatalf("expected missing-input error, got %v", err)
+	}
+	if _, err := os.Stat(filepath.Join(dir, "build")); !os.IsNotExist(err) {
+		t.Fatal("missing input must not create output dir")
+	}
+	if _, err := os.Stat(filepath.Join(dir, ".relspec")); !os.IsNotExist(err) {
+		t.Fatal("missing input must not create logfile")
+	}
+}
+
+func TestJobRun_MissingConnEnvNoExecution(t *testing.T) {
+	manifest := `version: 1
+jobs:
+  remote:
+    command: convert
+    inputs:
+      - format: pgsql
+        conn_env: RELSPEC_TEST_MISSING_CONN
+    output:
+      format: json
+      path: build/out.json
+    logfile: .relspec/remote.log
+`
+	dir := jobFixture(t, manifest)
+	os.Unsetenv("RELSPEC_TEST_MISSING_CONN")
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	err := executeJobPlan(set, "remote", false, false, &bytes.Buffer{})
+	if err == nil || !strings.Contains(err.Error(), "conn_env") {
+		t.Fatalf("expected missing conn_env error, got %v", err)
+	}
+	if _, err := os.Stat(filepath.Join(dir, ".relspec")); !os.IsNotExist(err) {
+		t.Fatal("missing conn_env must not create logfile")
+	}
+}
+
+func TestJobRun_ExitCodePropagation(t *testing.T) {
+	// gorm output without options.package makes the underlying writer fail.
+	manifest := `version: 1
+jobs:
+  fail:
+    command: convert
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    output:
+      format: gorm
+      path: build/models
+      overwrite: true
+    logfile: .relspec/fail.log
+`
+	dir := jobFixture(t, manifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	err := executeJobPlan(set, "fail", false, false, &bytes.Buffer{})
+	if err == nil {
+		t.Fatal("expected underlying failure to propagate")
+	}
+	if !strings.Contains(err.Error(), "job \"fail\" failed") {
+		t.Fatalf("error should identify the failing job: %v", err)
+	}
+	// Logfile records the failure and no misleading success marker.
+	logData, _ := os.ReadFile(filepath.Join(dir, ".relspec", "fail.log"))
+	if strings.Contains(string(logData), "\nOK\n") || strings.HasSuffix(strings.TrimSpace(string(logData)), "OK") {
+		t.Fatalf("failed job must not log OK:\n%s", logData)
+	}
+	if !strings.Contains(string(logData), "FAILED") {
+		t.Fatalf("failed job should log FAILED:\n%s", logData)
+	}
+}
+
+func TestJobRun_DependencyChainExecutes(t *testing.T) {
+	manifest := `version: 1
+jobs:
+  a:
+    command: convert
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/a.json
+      overwrite: true
+  b:
+    command: convert
+    depends_on: [a]
+    inputs:
+      - path: schema/tenant.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/b.json
+      overwrite: true
+`
+	dir := jobFixture(t, manifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	if err := executeJobPlan(set, "b", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("executeJobPlan: %v", err)
+	}
+	for _, f := range []string{"a.json", "b.json"} {
+		if _, err := os.Stat(filepath.Join(dir, "build", f)); err != nil {
+			t.Fatalf("expected %s to be produced: %v", f, err)
+		}
+	}
+}
+
+func TestJobRun_ScriptsListMultipleDirs(t *testing.T) {
+	dir := t.TempDir()
+	writeFile(t, filepath.Join(dir, "migrations", "core", "1_001_create_users.sql"), "CREATE TABLE users();\n")
+	writeFile(t, filepath.Join(dir, "migrations", "tenant", "1_002_create_posts.sql"), "CREATE TABLE posts();\n")
+	writeFile(t, filepath.Join(dir, "migrations", "tenant", "2_001_add_index.sql"), "CREATE INDEX x ON posts(id);\n")
+	manifest := `version: 1
+jobs:
+  list-all:
+    command: scripts-list
+    script_dirs:
+      - migrations/core
+      - migrations/tenant
+    logfile: .relspec/scripts.log
+`
+	writeFile(t, filepath.Join(dir, "relspec.yml"), manifest)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	if err := executeJobPlan(set, "list-all", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("executeJobPlan: %v", err)
+	}
+	logData, err := os.ReadFile(filepath.Join(dir, ".relspec", "scripts.log"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	s := string(logData)
+	iUsers := strings.Index(s, "create_users")
+	iPosts := strings.Index(s, "create_posts")
+	iIndex := strings.Index(s, "add_index")
+	if iUsers < 0 || iPosts < 0 || iIndex < 0 {
+		t.Fatalf("expected all scripts listed:\n%s", s)
+	}
+	if !(iUsers < iPosts && iPosts < iIndex) {
+		t.Fatalf("scripts not in priority/sequence order:\n%s", s)
+	}
+	if !strings.Contains(s, "found 3 script(s) across 2") {
+		t.Fatalf("expected multi-directory summary:\n%s", s)
+	}
+}
+
+func TestJobRun_ConnEnvRedactedInPlan(t *testing.T) {
+	manifest := `version: 1
+jobs:
+  remote:
+    command: convert
+    inputs:
+      - format: pgsql
+        conn_env: RELSPEC_TEST_PLAN_CONN
+    output:
+      format: json
+      path: build/out.json
+`
+	dir := jobFixture(t, manifest)
+	secret := "postgres://user:supersecret@db.example/app"
+	t.Setenv("RELSPEC_TEST_PLAN_CONN", secret)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+
+	var buf bytes.Buffer
+	if err := executeJobPlan(set, "remote", true, false, &buf); err != nil {
+		t.Fatalf("dry run: %v", err)
+	}
+	if strings.Contains(buf.String(), "supersecret") || strings.Contains(buf.String(), secret) {
+		t.Fatalf("plan leaked secret:\n%s", buf.String())
+	}
+	if !strings.Contains(buf.String(), "env:RELSPEC_TEST_PLAN_CONN") {
+		t.Fatalf("plan should reference the env var name:\n%s", buf.String())
+	}
+}
+
+func TestJobLogger_Redaction(t *testing.T) {
+	lg := &jobLogger{secrets: []string{"topsecret"}}
+	got := lg.redact("connecting with password topsecret and postgres://u:p@h/db")
+	if strings.Contains(got, "topsecret") {
+		t.Fatalf("secret not redacted: %q", got)
+	}
+	if !strings.Contains(got, "***") {
+		t.Fatalf("expected redaction marker: %q", got)
+	}
+}
+
+func TestJobList_DeterministicOutput(t *testing.T) {
+	manifest := `version: 1
+jobs:
+  zebra:
+    command: convert
+    inputs: [{path: schema/core.dbml, format: dbml}]
+    output: {format: json, path: build/z.json}
+  alpha:
+    command: convert
+    inputs: [{path: schema/core.dbml, format: dbml}]
+    output: {format: json, path: build/a.json}
+`
+	dir := jobFixture(t, manifest)
+
+	run := func() string {
+		jobDir = dir
+		jobFiles = nil
+		cmd := &cobra.Command{}
+		var buf bytes.Buffer
+		cmd.SetOut(&buf)
+		if err := runJobList(cmd, nil); err != nil {
+			t.Fatalf("runJobList: %v", err)
+		}
+		return buf.String()
+	}
+	first := run()
+	if strings.Index(first, "alpha") > strings.Index(first, "zebra") {
+		t.Fatalf("jobs not sorted:\n%s", first)
+	}
+	if first != run() {
+		t.Fatal("job list output not deterministic")
+	}
+}
+
+func TestJobRun_SplitJob(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  extract:
+    command: split
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    select:
+      tables: [users]
+    output:
+      format: json
+      path: build/subset.json
+      overwrite: true
+`)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "extract", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute split job: %v", err)
+	}
+	out, err := os.ReadFile(filepath.Join(dir, "build", "subset.json"))
+	if err != nil {
+		t.Fatalf("read split output: %v", err)
+	}
+	s := string(out)
+	if !strings.Contains(s, "users") {
+		t.Fatalf("split output missing selected table:\n%s", s)
+	}
+	if strings.Contains(s, "posts") {
+		t.Fatalf("split output should have excluded posts:\n%s", s)
+	}
+}
+
+func TestJobRun_InspectJob(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  lint:
+    command: inspect
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    report:
+      format: json
+      path: build/report.json
+      overwrite: true
+    logfile: .relspec/lint.log
+`)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	// Default rules only warn, so the job succeeds.
+	if err := executeJobPlan(set, "lint", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute inspect job: %v", err)
+	}
+	if _, err := os.ReadFile(filepath.Join(dir, "build", "report.json")); err != nil {
+		t.Fatalf("expected report file: %v", err)
+	}
+	logData, _ := os.ReadFile(filepath.Join(dir, ".relspec", "lint.log"))
+	if !strings.Contains(string(logData), "inspect:") {
+		t.Fatalf("logfile missing inspect summary:\n%s", logData)
+	}
+}
+
+func TestJobRun_InspectJobFailsOnRuleError(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  lint:
+    command: inspect
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    rules: rules.yaml
+    report:
+      format: json
+      path: build/report.json
+      overwrite: true
+    logfile: .relspec/lint.log
+`)
+	// A rule set to "error" level for a violation the fixture triggers.
+	writeFile(t, filepath.Join(dir, "rules.yaml"), `version: "1.0"
+rules:
+  primary_key_naming:
+    enabled: enforce
+    function: primary_key_naming
+    pattern: "^id_"
+    message: "Primary key columns should start with id_"
+`)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	err := executeJobPlan(set, "lint", false, false, &bytes.Buffer{})
+	if err == nil || !strings.Contains(err.Error(), "error(s)") {
+		t.Fatalf("expected inspect job to fail on rule error, got %v", err)
+	}
+	logData, _ := os.ReadFile(filepath.Join(dir, ".relspec", "lint.log"))
+	if !strings.Contains(string(logData), "FAILED") {
+		t.Fatalf("failed inspect job should log FAILED:\n%s", logData)
+	}
+}
+
+func TestJobRun_DiffJob(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  compare:
+    command: diff
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    report:
+      format: json
+      path: build/diff.json
+      overwrite: true
+`)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "compare", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute diff job: %v", err)
+	}
+	out, err := os.ReadFile(filepath.Join(dir, "build", "diff.json"))
+	if err != nil {
+		t.Fatalf("read diff report: %v", err)
+	}
+	if len(out) == 0 {
+		t.Fatal("diff report is empty")
+	}
+}
+
+func TestJobRun_FromJobWiring(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  a:
+    command: convert
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/a.json
+      overwrite: true
+  b:
+    command: convert
+    inputs:
+      - from_job: a
+    output:
+      format: yaml
+      path: build/b.yaml
+      overwrite: true
+`)
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "b", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute from_job chain: %v", err)
+	}
+	if _, err := os.Stat(filepath.Join(dir, "build", "a.json")); err != nil {
+		t.Fatalf("producer output missing: %v", err)
+	}
+	out, err := os.ReadFile(filepath.Join(dir, "build", "b.yaml"))
+	if err != nil {
+		t.Fatalf("consumer output missing: %v", err)
+	}
+	if !strings.Contains(string(out), "users") {
+		t.Fatalf("consumer did not consume producer output:\n%s", out)
+	}
+}
+
+func TestJobRun_LogRotation(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  build:
+    command: scripts-list
+    script_dirs: [migrations]
+    log_max_size: "150B"
+    log_keep: 2
+    logfile: .relspec/build.log
+`)
+	writeFile(t, filepath.Join(dir, "migrations", "1_001_a.sql"), "CREATE TABLE a();\n")
+	logPath := filepath.Join(dir, ".relspec", "build.log")
+	writeFile(t, logPath, strings.Repeat("x", 300)+"\n")
+
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "build", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute job: %v", err)
+	}
+	rotated, err := os.ReadFile(logPath + ".1")
+	if err != nil {
+		t.Fatalf("expected rotated logfile build.log.1: %v", err)
+	}
+	if !strings.Contains(string(rotated), strings.Repeat("x", 300)) {
+		t.Fatalf("rotated logfile should hold the old content")
+	}
+	fresh, err := os.ReadFile(logPath)
+	if err != nil {
+		t.Fatalf("expected fresh logfile: %v", err)
+	}
+	if strings.Contains(string(fresh), strings.Repeat("x", 300)) {
+		t.Fatalf("fresh logfile should not contain the rotated-out content:\n%s", fresh)
+	}
+	if !strings.Contains(string(fresh), "OK") {
+		t.Fatalf("fresh logfile should hold the new run:\n%s", fresh)
+	}
+}
+
+func TestJobRun_AtomicOutputLeavesOriginalOnFailure(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  x:
+    command: convert
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/out.json
+      overwrite: true
+`)
+	// Seed the destination, then make its parent directory read-only so the
+	// rename step fails. The seeded file must survive intact.
+	seeded := filepath.Join(dir, "build", "out.json")
+	writeFile(t, seeded, `{"seeded":true}`)
+	if err := os.Chmod(filepath.Join(dir, "build"), 0o500); err != nil {
+		t.Skipf("cannot chmod: %v", err)
+	}
+	t.Cleanup(func() { _ = os.Chmod(filepath.Join(dir, "build"), 0o755) })
+
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "x", false, false, &bytes.Buffer{}); err == nil {
+		t.Skip("write unexpectedly succeeded (running as root?)")
+	}
+	if err := os.Chmod(filepath.Join(dir, "build"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	data, err := os.ReadFile(seeded)
+	if err != nil {
+		t.Fatalf("seeded file gone: %v", err)
+	}
+	if !strings.Contains(string(data), "seeded") {
+		t.Fatalf("seeded file was corrupted: %s", data)
+	}
+}
+
+func TestJobRun_ScriptsExecMissingConnEnv(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  migrate:
+    command: scripts-exec
+    script_dirs: [migrations]
+    output:
+      conn_env: RELSPEC_TEST_EXEC_MISSING
+    logfile: .relspec/migrate.log
+`)
+	writeFile(t, filepath.Join(dir, "migrations", "1_001_a.sql"), "CREATE TABLE a();\n")
+	os.Unsetenv("RELSPEC_TEST_EXEC_MISSING")
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	err := executeJobPlan(set, "migrate", false, false, &bytes.Buffer{})
+	if err == nil || !strings.Contains(err.Error(), "conn_env") {
+		t.Fatalf("expected missing conn_env error, got %v", err)
+	}
+}
+
+func TestJobRun_ScriptsExecDryRun(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  migrate:
+    command: scripts-exec
+    script_dirs: [migrations]
+    output:
+      conn_env: RELSPEC_TEST_EXEC_CONN
+`)
+	writeFile(t, filepath.Join(dir, "migrations", "1_001_a.sql"), "CREATE TABLE a();\n")
+	t.Setenv("RELSPEC_TEST_EXEC_CONN", "postgres://u:secretpw@h/db")
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	var buf bytes.Buffer
+	if err := executeJobPlan(set, "migrate", true, false, &buf); err != nil {
+		t.Fatalf("dry run: %v", err)
+	}
+	if strings.Contains(buf.String(), "secretpw") {
+		t.Fatalf("plan leaked secret:\n%s", buf.String())
+	}
+	if !strings.Contains(buf.String(), "env:RELSPEC_TEST_EXEC_CONN") {
+		t.Fatalf("plan should name the env var:\n%s", buf.String())
+	}
+}
+
+func TestJobRun_TemplDatabaseMode(t *testing.T) {
+	dir := jobFixture(t, `version: 1
+jobs:
+  docs:
+    command: templ
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+    template: templates/schema.tmpl
+    output:
+      path: build/schema.txt
+      overwrite: true
+`)
+	writeFile(t, filepath.Join(dir, "templates", "schema.tmpl"), "{{range .Database.Schemas}}{{range .Tables}}{{.Name}} {{end}}{{end}}")
+	set := mustLoadSet(t, filepath.Join(dir, "relspec.yml"))
+	if err := executeJobPlan(set, "docs", false, false, &bytes.Buffer{}); err != nil {
+		t.Fatalf("execute templ job: %v", err)
+	}
+	out, err := os.ReadFile(filepath.Join(dir, "build", "schema.txt"))
+	if err != nil {
+		t.Fatalf("read templ output: %v", err)
+	}
+	if !strings.Contains(string(out), "users") {
+		t.Fatalf("templ output missing users table: %s", out)
+	}
+}

cmd/relspec/main.go

@@ -6,6 +6,7 @@ import (
 )
 
 func main() {
+	printVersionHeader(os.Args[1:])
 	if err := rootCmd.Execute(); err != nil {
 		fmt.Fprintln(os.Stderr, err)
 		os.Exit(1)

cmd/relspec/merge.go

@@ -158,9 +158,7 @@ func runMerge(cmd *cobra.Command, args []string) error {
 		}
 		mergeTargetPath = expandPath(mergeTargetPath)
 	} else if mergeTargetConn == "" {
-
 		return fmt.Errorf("--target-conn is required for pgsql format")
-
 	}
 
 	if mergeSourceType != "pgsql" {

cmd/relspec/merge_from_list_test.go

@@ -105,7 +105,7 @@ func TestRunMerge_FromListPathWithSpaces(t *testing.T) {
 	defer restoreMergeState(saved)
 
 	spacedDir := filepath.Join(t.TempDir(), "my schema files")
-	if err := os.MkdirAll(spacedDir, 0755); err != nil {
+	if err := os.MkdirAll(spacedDir, 0o755); err != nil {
 		t.Fatal(err)
 	}
 	targetFile := filepath.Join(spacedDir, "target schema.json")

cmd/relspec/prisma_options.go

@@ -10,6 +10,7 @@ func newReaderOptions(filePath, connString string) *readers.ReaderOptions {
 		FilePath:         filePath,
 		ConnectionString: connString,
 		Prisma7:          prisma7,
+		StrictDirectives: strictDirectives,
 	}
 }
 
@@ -22,5 +23,6 @@ func newWriterOptions(outputPath, packageName string, flattenSchema bool, nullab
 		NullableArrays:   nullableArrays,
 		Prisma7:          prisma7,
 		ContinueOnError:  continueOnError,
+		StrictDirectives: strictDirectives,
 	}
 }

cmd/relspec/root.go

@@ -13,6 +13,8 @@ var (
 	version          = "dev"
 	buildDate        = "unknown"
 	prisma7          bool
+	noVersion        bool
+	strictDirectives bool
 )
 
 func init() {
@@ -54,9 +56,6 @@ bidirectional conversion between various database schema formats.
 It reads database schemas from multiple sources (live databases, DBML,
 DCTX, DrawDB, etc.) and writes them to various formats (GORM, Bun,
 JSON, YAML, SQL, etc.).`,
-	PersistentPreRun: func(cmd *cobra.Command, args []string) {
-		fmt.Printf("RelSpec %s (built: %s)\n\n", version, buildDate)
-	},
 }
 
 func init() {
@@ -64,6 +63,7 @@ func init() {
 	rootCmd.AddCommand(diffCmd)
 	rootCmd.AddCommand(inspectCmd)
 	rootCmd.AddCommand(scriptsCmd)
+	rootCmd.AddCommand(jobCmd)
 	rootCmd.AddCommand(assetsCmd)
 	rootCmd.AddCommand(templCmd)
 	rootCmd.AddCommand(editCmd)
@@ -72,4 +72,23 @@ func init() {
 	rootCmd.AddCommand(versionCmd)
 	rootCmd.AddCommand(reportCmd)
 	rootCmd.PersistentFlags().BoolVar(&prisma7, "prisma7", false, "Use Prisma 7 generator conventions when reading/writing Prisma schemas")
+	rootCmd.PersistentFlags().BoolVar(&noVersion, "no-version", false, "Suppress the RelSpec version header")
+	rootCmd.PersistentFlags().BoolVar(&strictDirectives, "strict-directives", false, "Fail on unknown or untranslatable DBML dialect directives (@postgres:, @sqlite:, …)")
+}
+
+// printVersionHeader prints the "RelSpec <version> (built: <date>)" banner
+// that precedes all command output. It is invoked from main() before cobra
+// parses/executes anything, so it runs even for --help and bare invocations.
+// It is skipped when --no-version is present, or when the version subcommand
+// is being run (which prints its own, more detailed output).
+func printVersionHeader(args []string) {
+	for _, a := range args {
+		if a == "--no-version" {
+			return
+		}
+	}
+	if len(args) > 0 && args[0] == "version" {
+		return
+	}
+	fmt.Printf("RelSpec %s (built: %s)\n\n", version, buildDate)
 }

cmd/relspec/split.go

@@ -205,8 +205,52 @@ func runSplit(cmd *cobra.Command, args []string) error {
 	return nil
 }
 
-// filterDatabase filters the database based on provided criteria
+// splitSelection is the schema/table selection for a split, independent of the
+// CLI flag globals so the job runner can build one directly.
+type splitSelection struct {
+	Schemas        []string
+	Tables         []string
+	ExcludeSchemas []string
+	ExcludeTables  []string
+	DatabaseName   string
+}
+
+// summary renders a one-line human description of the selection.
+func (s splitSelection) summary() string {
+	var parts []string
+	if len(s.Schemas) > 0 {
+		parts = append(parts, "schemas="+strings.Join(s.Schemas, ","))
+	}
+	if len(s.Tables) > 0 {
+		parts = append(parts, "tables="+strings.Join(s.Tables, ","))
+	}
+	if len(s.ExcludeSchemas) > 0 {
+		parts = append(parts, "exclude_schemas="+strings.Join(s.ExcludeSchemas, ","))
+	}
+	if len(s.ExcludeTables) > 0 {
+		parts = append(parts, "exclude_tables="+strings.Join(s.ExcludeTables, ","))
+	}
+	if s.DatabaseName != "" {
+		parts = append(parts, "database_name="+s.DatabaseName)
+	}
+	if len(parts) == 0 {
+		return "(all schemas/tables)"
+	}
+	return strings.Join(parts, " ")
+}
+
+// filterDatabase filters the database based on the CLI split flags.
 func filterDatabase(db *models.Database) (*models.Database, error) {
+	return filterDatabaseSelection(db, splitSelection{
+		Schemas:        parseCommaSeparated(splitSchemas),
+		Tables:         parseCommaSeparated(splitTables),
+		ExcludeSchemas: parseCommaSeparated(splitExcludeSchema),
+		ExcludeTables:  parseCommaSeparated(splitExcludeTables),
+	})
+}
+
+// filterDatabaseSelection filters db down to the schemas/tables named by sel.
+func filterDatabaseSelection(db *models.Database, sel splitSelection) (*models.Database, error) {
 	filteredDB := &models.Database{
 		Name:            db.Name,
 		Description:     db.Description,
@@ -220,11 +264,11 @@ func filterDatabase(db *models.Database) (*models.Database, error) {
 		Domains:         db.Domains, // Keep domains for now
 	}
 
-	// Parse filter flags
-	includeSchemas := parseCommaSeparated(splitSchemas)
-	includeTables := parseCommaSeparated(splitTables)
-	excludeSchemas := parseCommaSeparated(splitExcludeSchema)
-	excludeTables := parseCommaSeparated(splitExcludeTables)
+	// Selection criteria
+	includeSchemas := sel.Schemas
+	includeTables := sel.Tables
+	excludeSchemas := sel.ExcludeSchemas
+	excludeTables := sel.ExcludeTables
 
 	// Convert table names to lowercase for case-insensitive matching
 	includeTablesLower := make(map[string]bool)

cmd/relspec/templ_from_list_test.go

@@ -10,7 +10,7 @@ import (
 func writeTestTemplate(t *testing.T, path string) {
 	t.Helper()
 	content := []byte(`{{.Name}}`)
-	if err := os.WriteFile(path, content, 0644); err != nil {
+	if err := os.WriteFile(path, content, 0o644); err != nil {
 		t.Fatalf("failed to write template file %s: %v", path, err)
 	}
 }
@@ -104,7 +104,7 @@ func TestRunTempl_FromListPathWithSpaces(t *testing.T) {
 	defer restoreTemplState(saved)
 
 	spacedDir := filepath.Join(t.TempDir(), "my schema files")
-	if err := os.MkdirAll(spacedDir, 0755); err != nil {
+	if err := os.MkdirAll(spacedDir, 0o755); err != nil {
 		t.Fatal(err)
 	}
 	file1 := filepath.Join(spacedDir, "users schema.json")

cmd/relspec/testhelpers_test.go

@@ -66,7 +66,7 @@ func writeTestJSON(t *testing.T, path string, tableNames []string) {
 	if err != nil {
 		t.Fatalf("failed to marshal test JSON: %v", err)
 	}
-	if err := os.WriteFile(path, data, 0644); err != nil {
+	if err := os.WriteFile(path, data, 0o644); err != nil {
 		t.Fatalf("failed to write test file %s: %v", path, err)
 	}
 }
@@ -100,7 +100,7 @@ func writeTestJSONWithSingleColumnType(t *testing.T, path, tableName, columnType
 	if err != nil {
 		t.Fatalf("failed to marshal test JSON: %v", err)
 	}
-	if err := os.WriteFile(path, data, 0644); err != nil {
+	if err := os.WriteFile(path, data, 0o644); err != nil {
 		t.Fatalf("failed to write test file %s: %v", path, err)
 	}
 }

docs/DBML_DIRECTIVES.md

@@ -0,0 +1,115 @@
+# DBML Dialect Directives
+
+DBML has no dialect-neutral way to express database-specific features such as
+PostgreSQL table partitioning or SQLite `WITHOUT ROWID`. RelSpec adds **dialect
+directives** — explicit, parseable lines embedded in a `.dbml` file that are:
+
+- stored losslessly in the intermediate model (under each object's `Metadata`),
+- preserved unchanged through a `DBML → model → DBML` round-trip,
+- translated to SQL **only** by the writer for the matching dialect
+  (`@postgres:` clauses appear in PostgreSQL output, never in SQLite output, and
+  vice-versa).
+
+## Grammar
+
+A directive is a single line, matched on its trimmed content:
+
+```
+@<namespace>[(<target>)]: <args>
+```
+
+| Part | Rules |
+|------|-------|
+| `namespace` | `^[a-z][a-z0-9_]*$` — e.g. `postgres`, `sqlite`. Future dialects allowed. |
+| `(target)` | Optional. A **column name** only, valid only on a directive line inside a table body. Bare or single/double quoted. |
+| `args` | Everything after the first `:`, trimmed. Otherwise preserved **verbatim**. Must be non-empty. |
+
+The **key** of a directive is derived: the lowercased first whitespace-delimited
+token of `args` (`partition by RANGE (created_at)` → `partition`). It drives
+duplicate detection and writer dispatch.
+
+## Location
+
+Where the line appears determines which object it attaches to:
+
+| Position in the file | Attaches to |
+|----------------------|-------------|
+| Before the first `Table {` | database (`db.Metadata`) |
+| Table body, no `(target)` | that table |
+| Table body, `(col)` target | column `col` of that table (error if `col` is unknown) |
+| Inside an `indexes { }` block | the **most recently listed** index entry in that block; `(target)` is not allowed |
+
+```dbml
+@postgres: search_path myapp          -- database
+
+Table myapp.events {
+  id bigint [pk]
+  created_at timestamp [not null]
+  @postgres(id): identity always      -- column "id"
+  @postgres: partition by RANGE (created_at)   -- table
+  @postgres: tablespace fast_data              -- table
+  @sqlite: without rowid                        -- table
+
+  indexes {
+    (created_at) [name: 'idx_events_created']
+    @postgres: with (fillfactor=90)   -- index "idx_events_created"
+    @postgres: tablespace idx_space    -- index "idx_events_created"
+  }
+}
+```
+
+## Duplicate policy
+
+- **Repeatable by default** — every directive with the same `(namespace, key)` at
+  one location is kept, in source order.
+- **Singletons** raise a line-numbered error on a second occurrence at the same
+  location. Current singletons: `postgres` `partition`, `tablespace`, `inherits`,
+  `storage`, `compression`, `identity`; `sqlite` `without`, `strict`, `collate`.
+
+## Strict mode
+
+CLI flag `--strict-directives` (also `ReaderOptions.StrictDirectives` /
+`WriterOptions.StrictDirectives`):
+
+- **Reader**: an unknown namespace or key is a hard error. Without strict mode it
+  is stored and preserved silently, and round-trips unchanged.
+- **PostgreSQL / SQLite writer**: a directive for **that** writer's own dialect
+  whose key it cannot translate is a hard error. Without strict mode, translatable
+  keys are emitted and the rest are skipped. Directives for other dialects are
+  always ignored, never emitted.
+
+## Errors
+
+All are line-numbered (`dbml: line N: …`):
+
+- no colon, or empty `args`
+- namespace empty or not matching `[a-z][a-z0-9_]*`
+- `(target)` naming an unknown column, or used at the top level / in an `indexes` block
+- a directive in the catalog used at a location it is not valid for
+- duplicate singleton at the same location
+- (strict mode) unknown `(namespace, key)`
+
+## Supported directive matrix
+
+### `@postgres`
+
+| Key | Locations | SQL emitted | Notes |
+|-----|-----------|-------------|-------|
+| `partition` | table | `PARTITION BY <args>` appended to `CREATE TABLE` | e.g. `@postgres: partition by RANGE (created_at)` |
+| `inherits` | table | `INHERITS (<args>)` — args verbatim | |
+| `with` | table, index | `WITH (<params>)` | On an index, wins over `WITH` derived from the index comment. `@postgres: with (fillfactor=90)` |
+| `tablespace` | table, index | `TABLESPACE <name>` | Emitted after `WITH`, before `WHERE` on indexes |
+| `storage` | column | `STORAGE <mode>` in the column definition | e.g. `@postgres(blob): storage external` |
+| `compression` | column | `COMPRESSION <method>` | |
+| `identity` | column | `identity always` → `GENERATED ALWAYS AS IDENTITY`; `identity default` / `identity by default` → `GENERATED BY DEFAULT AS IDENTITY` | |
+
+### `@sqlite`
+
+| Key | Locations | SQL emitted | Notes |
+|-----|-----------|-------------|-------|
+| `without` | table | `WITHOUT ROWID` table option | `@sqlite: without rowid` |
+| `strict` | table | `STRICT` table option | `WITHOUT ROWID` is emitted before `STRICT` |
+| `collate` | column | ` COLLATE <name>` in the column definition | e.g. `@sqlite(name): collate NOCASE` |
+
+Unknown namespaces and keys not in these tables are still preserved losslessly
+(and round-trip through the DBML writer) whenever strict mode is off.

docs/JOB_FILES.md

@@ -0,0 +1,365 @@
+# RelSpec Job Files
+
+Job files let you declare named, repeatable RelSpec workflows in YAML and run
+them with `relspec job run <name>` instead of retyping long command lines.
+
+```bash
+relspec job list                     # deterministic list of discovered jobs
+relspec job run build-schema --plan  # validate + print plan, execute nothing
+relspec job run build-schema         # run the job (and its dependencies)
+```
+
+## Design contract
+
+This is a deliberately small, safe contract. Every capability is offline-testable
+except live database execution (`scripts-exec`), which is validated and planned
+offline and only connects at run time.
+
+### Not a shell
+
+`command` is a **closed allow-list**. There is no field anywhere that accepts a
+shell string, an executable path, or arbitrary arguments. Adding a new command
+means adding a vetted adapter in the RelSpec source.
+
+| command        | what it does                                                        |
+|----------------|--------------------------------------------------------------------|
+| `convert`      | read one or more input schemas, additively merge them, write one output |
+| `merge`        | like `convert` but requires ≥2 inputs and exposes `skip_*` merge options |
+| `split`        | read one or more schemas, keep the selected schemas/tables, write one output |
+| `scripts-list` | deterministically list SQL scripts across one or more directories   |
+| `scripts-exec` | execute SQL scripts across one or more directories against a live PostgreSQL database |
+| `templ`        | apply a custom Go text template to one or more input schemas          |
+| `inspect`      | validate one or more schemas against rules and write a report       |
+| `diff`         | compare exactly two schemas and write a differences report          |
+
+`convert`, `merge` and `split` are **producers**: their file output can be fed
+directly into another job with `from_job` (see below).
+
+### Discovery and precedence
+
+`relspec job` (no `--file`) scans `--dir` (default `.`) for:
+
+1. `relspec.yml` / `relspec.yaml` (the default file), then
+2. `relspec.<name>.yml` / `relspec.<name>.yaml` (extra files),
+
+each group sorted lexically. Order is stable across runs. Use `--file <path>`
+(repeatable) to load explicit files and skip discovery.
+
+All discovered/selected files are merged into one job namespace. A job name
+defined by **more than one file is a hard error** naming both files. YAML maps
+already forbid duplicate keys within a single file.
+
+### Paths
+
+* Every path (`inputs[].path`, `output.path`, `report.path`, `rules`,
+  `script_dirs[]`, `template`, `logfile`) is **relative to the directory
+  containing the job file that declared the job**, not the process working
+  directory.
+* Absolute paths, `~`-relative paths and any path that resolves outside the job
+  file directory (`../`, `a/../../b`, …) are **rejected during validation** —
+  before anything runs.
+* At run time each path is additionally resolved through its symlinks: a symlink
+  inside the job-file directory that points outside it is rejected before the
+  path is opened.
+
+### Credentials
+
+* Database inputs (`format: pgsql` / `mssql`) and database execution outputs
+  (`format: pgsql` with `conn_env`) reference an **environment variable name**
+  via `conn_env:`. The connection string itself is never stored in the
+  manifest.
+* A `conn_env` value that looks like a connection string (contains `:`, `/`,
+  `@`, `=`, spaces) is rejected.
+* Missing/empty environment variables are reported during pre-flight, before
+  execution.
+* Job logs and `--plan` output show `env:<NAME>`, never the value. Resolved
+  secret values and anything matching a connection-string password are
+  redacted (`***`) from the logfile and diagnostics.
+
+### Validation happens before execution
+
+`relspec job list` and `relspec job run` both fully validate the selected set
+first. Nothing is read, written, connected to, or executed if validation fails.
+Checks include:
+
+* schema `version` — **forward-permissive**: any version `>= 1` is accepted.
+  An omitted `version` is treated as the current one. A version newer than this
+  build understands loads best-effort (unknown YAML fields are ignored and a
+  warning is printed); at the current version unknown YAML fields are still
+  rejected.
+* duplicate job names across files
+* unknown / missing `command`
+* per-command input/output shape:
+  * `convert` needs ≥1 input + output; `merge` needs ≥2 inputs + output
+  * `split` needs ≥1 input + a file output, plus an optional `select:` block
+  * `scripts-list` needs `script_dirs` and forbids inputs/output
+  * `scripts-exec` needs `script_dirs` and `output.conn_env` (pgsql only)
+  * `inspect` needs ≥1 input + `report:` (format `markdown`|`json`)
+  * `diff` needs **exactly 2** inputs + `report:` (format `summary`|`json`|`html`)
+* unknown input/output `format`
+* `from_job` targets exist, are producers (`convert`/`merge`/`split`) and write a
+  single-file output
+* path traversal / absolute / home-relative paths
+* `depends_on` and `from_job` targets exist
+* dependency cycles over the combined `depends_on` + `from_job` graph
+  (reported as `a -> b -> c -> a`)
+
+Then, immediately before running, per-job pre-flight resolves paths and checks:
+
+* every input file exists and is a file (a `from_job` input is exempt — its
+  producer runs earlier in the same plan)
+* every `script_dir` exists and is a directory
+* every `conn_env` variable is set
+* `output.path` / `report.path` does not already exist unless the matching
+  `overwrite: true` is set
+* `rules` (inspect), when given, exists and is a file
+* symlinks in every resolved path stay inside the job-file directory
+
+If any pre-flight check fails for **any** job in the plan, **no** job runs.
+
+### Execution and exit codes
+
+* `relspec job run <name>` runs the job's dependency closure first
+  (`depends_on` plus any `from_job` producers), in topological order
+  (deterministic), then the job. `--no-deps` runs only the named job and is
+  incompatible with `from_job` inputs.
+* `--dry-run` (alias `--plan`) prints the resolved plan and exits 0 without
+  touching inputs, outputs or databases.
+* A failing job returns the underlying non-zero status (the process exits 1)
+  and the error names the job. The logfile records `FAILED: <error>`; a
+  successful job records `OK`. No separate success-marker file is written, so a
+  failure can never leave a stale "success".
+* `inspect` fails the job when the report contains rule **errors** (enforced
+  rules); warnings do not fail it. `diff` never fails on differences.
+* Single-file outputs and reports are written to a temporary file in the target
+  directory and atomically renamed into place, so an interrupted run never
+  leaves a partial file. Directory-emitting formats (`gorm`, `bun`, `drizzle`,
+  `typeorm`, `prisma`) are written in place.
+
+### Logfile rotation
+
+When a job has a `logfile`, it is size-rotated before each run. Defaults are
+**5 MB** with **3** rotated files kept (`build.log` → `build.log.1` → …). Override
+per job with `log_max_size` / `log_keep`, or for a whole file with a top-level
+`defaults:` block. `log_max_size` accepts `B`/`KB`/`MB`/`GB` suffixes (e.g.
+`"512KB"`, `"5MB"`).
+
+## Schema reference
+
+```yaml
+version: 1                       # optional; any value >= 1 is accepted
+defaults:                        # optional, file-wide
+  log_max_size: 5MB              # B / KB / MB / GB
+  log_keep: 3
+jobs:
+  <job-name>:
+    command: convert | merge | split | scripts-list | scripts-exec | templ | inspect | diff
+    description: "free text"                   # optional, shown by `job list`
+    depends_on: [other-job, ...]               # optional
+    inputs:                                    # convert (≥1) / merge (≥2) / split (≥1) / inspect (≥1) / diff (exactly 2)
+      - path: relative/file.dbml               # file inputs
+        format: dbml
+      - format: pgsql                          # live-connection inputs
+        conn_env: SOURCE_DB_URL                # env var NAME
+      - from_job: build-schema                 # consume another job's file output
+    script_dirs:                               # scripts-list / scripts-exec (≥1)
+      - migrations/core
+      - migrations/tenant
+    template: templates/schema.tmpl             # templ (required)
+    mode: table                                 # templ: database/schema/script/table
+    filename_pattern: "{{.Name}}.go"           # templ multi-output modes
+    select:                                    # split (optional; default = keep everything)
+      schemas: [public]
+      tables: [users, orders]
+      exclude_schemas: []
+      exclude_tables: []
+      database_name: SubsetDB                  # optional rename of the output database
+    rules: .relspec-rules.yaml                 # inspect (optional; built-in defaults if omitted)
+    report:                                    # inspect (required) / diff (required)
+      format: json                             # inspect: markdown|json ; diff: summary|json|html
+      path: build/report.json                  # required, except a diff "summary" (goes to the log)
+      overwrite: false
+    output:                                    # convert / merge / split (required); scripts-exec (required, conn_env)
+      format: pgsql
+      path: build/schema.sql                   # file output, OR:
+      conn_env: TARGET_DB_URL                  # execute against DB (pgsql only)
+      overwrite: false                         # default false
+    options:
+      flatten_schema: false
+      schema: public
+      package: models                          # for gorm/bun output
+      continue_on_error: false                 # pgsql / scripts-exec output
+      skip_relations: false                    # merge only
+      skip_enums: false
+      skip_views: false
+      skip_domains: false
+      skip_sequences: false
+    logfile: .relspec/log/<job-name>.log       # optional; appended to, size-rotated
+    log_max_size: 5MB                           # optional per-job override
+    log_keep: 3                                 # optional per-job override
+```
+
+For `templ`, `inputs` use the same file or `pgsql`/`conn_env` source forms as
+schema conversion. `output` is optional (empty means stdout); when present it
+contains only `path` and `overwrite`, because templates do not select a schema
+writer format.
+
+A `from_job` input takes no `path`, `format` or `conn_env`: it resolves to the
+named job's `output.path` and inherits its format, and implies a dependency on
+that job. The producer must be a `convert`, `merge` or `split` job writing a
+single-file output.
+
+### Supported input formats
+
+`dbml`, `dctx`, `drawdb`, `graphql`, `json`, `yaml`, `gorm`, `bun`, `drizzle`,
+`prisma`, `typeorm`, `sqlite` (file, via `path`); `pgsql`, `mssql`
+(live, via `conn_env`).
+
+### Supported output formats
+
+`dbml`, `dctx`, `drawdb`, `graphql`, `json`, `yaml`, `gorm`, `bun`, `drizzle`,
+`prisma`, `typeorm`, `pgsql`, `mssql`, `sqlite` (file, via `path`); `pgsql` also
+supports `conn_env` to execute the generated DDL against a live database.
+
+## Examples
+
+### Merge many schema files, emit PostgreSQL DDL
+
+```yaml
+version: 1
+jobs:
+  build-schema:
+    command: convert
+    inputs:
+      - { path: schema/core.dbml, format: dbml }
+      - { path: schema/billing.dbml, format: dbml }
+      - { path: schema/tenant.dbml, format: dbml }
+    output:
+      format: pgsql
+      path: build/schema.sql
+      overwrite: true
+    logfile: .relspec/log/build-schema.log
+```
+
+### Multiple script directories
+
+```yaml
+version: 1
+jobs:
+  migration-order:
+    command: scripts-list
+    script_dirs:
+      - migrations/core
+      - migrations/tenant
+      - migrations/reporting
+    logfile: .relspec/log/migration-order.log
+```
+
+### Job depending on another job
+
+```yaml
+version: 1
+jobs:
+  build-schema:
+    command: convert
+    inputs:
+      - { path: schema/core.dbml, format: dbml }
+      - { path: schema/tenant.dbml, format: dbml }
+    output: { format: json, path: build/schema.json, overwrite: true }
+  build-docs:
+    command: convert
+    depends_on: [build-schema]
+    inputs:
+      - { path: schema/core.dbml, format: dbml }
+    output: { format: yaml, path: build/schema.yaml, overwrite: true }
+```
+
+### Reading from a remote database
+
+```yaml
+version: 1
+jobs:
+  snapshot-prod:
+    command: convert
+    inputs:
+      - format: pgsql
+        conn_env: PROD_DB_URL      # export PROD_DB_URL=postgres://...
+    output:
+      format: dbml
+      path: snapshots/prod.dbml
+      overwrite: true
+```
+
+### Chain jobs with `from_job`, then lint the result
+
+```yaml
+version: 1
+jobs:
+  build-json:
+    command: convert
+    inputs:
+      - { path: schema/core.dbml, format: dbml }
+      - { path: schema/tenant.dbml, format: dbml }
+    output: { format: json, path: build/schema.json, overwrite: true }
+  lint-schema:
+    command: inspect
+    inputs:
+      - from_job: build-json      # implies depends_on: [build-json]
+    rules: .relspec-rules.yaml    # optional; built-in rules if omitted
+    report:
+      format: markdown
+      path: build/lint-report.md
+      overwrite: true
+```
+
+`relspec job run lint-schema` runs `build-json` first, then inspects its output.
+The job fails (exit 1) if any enforced rule is violated.
+
+### Split a subset out of a larger schema
+
+```yaml
+version: 1
+jobs:
+  posts-only:
+    command: split
+    inputs:
+      - { path: schema/core.dbml, format: dbml }
+      - { path: schema/tenant.dbml, format: dbml }
+    select:
+      tables: [posts]
+    output: { format: dbml, path: build/posts.dbml, overwrite: true }
+```
+
+### Diff two schemas
+
+```yaml
+version: 1
+jobs:
+  drift:
+    command: diff
+    inputs:                       # exactly two
+      - { path: build/schema.json, format: json }
+      - format: pgsql
+        conn_env: PROD_DB_URL
+    report:
+      format: summary             # summary → logfile; json/html need a path
+```
+
+`diff` reports differences and always exits 0.
+
+### Execute migration scripts against a live database
+
+```yaml
+version: 1
+jobs:
+  apply-migrations:
+    command: scripts-exec
+    script_dirs:
+      - migrations/core
+      - migrations/tenant
+    output:
+      conn_env: TARGET_DB_URL     # pgsql only; no path
+    options:
+      continue_on_error: false
+    logfile: .relspec/log/apply-migrations.log
+```

examples/jobs/.gitignore

@@ -0,0 +1,3 @@
+# Generated by `relspec job run` in this example project.
+/build/
+/.relspec/

examples/jobs/migrations/core/1_001_create_users.sql

@@ -0,0 +1,4 @@
+CREATE TABLE users (
+    id SERIAL PRIMARY KEY,
+    email VARCHAR NOT NULL UNIQUE
+);

examples/jobs/migrations/tenant/1_002_create_posts.sql

@@ -0,0 +1,5 @@
+CREATE TABLE posts (
+    id SERIAL PRIMARY KEY,
+    user_id INT NOT NULL REFERENCES users(id),
+    title VARCHAR NOT NULL
+);

examples/jobs/migrations/tenant/2_001_add_posts_user_idx.sql

@@ -0,0 +1 @@
+CREATE INDEX posts_user_id_idx ON posts(user_id);

examples/jobs/relspec.yml

@@ -0,0 +1,79 @@
+# Example RelSpec job file. See docs/JOB_FILES.md for the full reference.
+#
+#   cd examples/jobs
+#   relspec job list
+#   relspec job run build-schema --plan
+#   relspec job run build-schema
+#   relspec job run lint-schema        # inspect, consuming build-json's output
+version: 1
+
+# File-wide defaults. Individual jobs may override log_max_size / log_keep.
+defaults:
+  log_max_size: 2MB
+  log_keep: 5
+
+jobs:
+  build-schema:
+    command: convert
+    description: Merge the DBML sources and emit PostgreSQL DDL
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    output:
+      format: pgsql
+      path: build/schema.sql
+      overwrite: true
+    options:
+      flatten_schema: false
+    logfile: .relspec/log/build-schema.log
+
+  build-json:
+    command: convert
+    description: Also emit a JSON schema once build-schema succeeds
+    depends_on: [build-schema]
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    output:
+      format: json
+      path: build/schema.json
+      overwrite: true
+
+  migration-order:
+    command: scripts-list
+    description: Show the combined execution order across script directories
+    script_dirs:
+      - migrations/core
+      - migrations/tenant
+    logfile: .relspec/log/migration-order.log
+
+  lint-schema:
+    command: inspect
+    description: Validate build-json's output against the built-in rules
+    # No depends_on needed: the from_job input implies a dependency on build-json.
+    inputs:
+      - from_job: build-json
+    report:
+      format: markdown
+      path: build/lint-report.md
+      overwrite: true
+    logfile: .relspec/log/lint-schema.log
+
+  posts-only:
+    command: split
+    description: Extract just the posts table into its own DBML file
+    inputs:
+      - path: schema/core.dbml
+        format: dbml
+      - path: schema/tenant.dbml
+        format: dbml
+    select:
+      tables: [posts]
+    output:
+      format: dbml
+      path: build/posts.dbml
+      overwrite: true

examples/jobs/schema/core.dbml

@@ -0,0 +1,5 @@
+Table users {
+  id int [pk, increment]
+  email varchar [not null, unique]
+  created_at timestamp
+}

examples/jobs/schema/tenant.dbml

@@ -0,0 +1,6 @@
+Table posts {
+  id int [pk, increment]
+  user_id int [not null, ref: > users.id]
+  title varchar [not null]
+  body text
+}

go.mod

@@ -1,6 +1,6 @@
 module git.warky.dev/wdevs/relspecgo
 
-go 1.25.7
+go 1.25.13
 
 require (
 	github.com/gdamore/tcell/v2 v2.13.9
@@ -12,7 +12,7 @@ require (
 	github.com/stretchr/testify v1.11.1
 	github.com/uptrace/bun v1.2.18
 	github.com/uptrace/bun/dialect/pgdialect v1.2.18
-	golang.org/x/text v0.37.0
+	golang.org/x/text v0.39.0
 	gopkg.in/yaml.v3 v3.0.1
 	modernc.org/sqlite v1.50.1
 )
@@ -42,11 +42,11 @@ require (
 	github.com/tmthrgd/go-hex v0.0.0-20190904060850-447a3041c3bc // indirect
 	github.com/vmihailenco/msgpack/v5 v5.4.1 // indirect
 	github.com/vmihailenco/tagparser/v2 v2.0.0 // indirect
-	golang.org/x/crypto v0.51.0 // indirect
-	golang.org/x/sync v0.20.0 // indirect
-	golang.org/x/sys v0.44.0 // indirect
-	golang.org/x/term v0.43.0 // indirect
-	golang.org/x/tools v0.45.0 // indirect
+	golang.org/x/crypto v0.53.0 // indirect
+	golang.org/x/net v0.56.0 // indirect
+	golang.org/x/sync v0.21.0 // indirect
+	golang.org/x/sys v0.46.0 // indirect
+	golang.org/x/term v0.44.0 // indirect
 	modernc.org/libc v1.72.3 // indirect
 	modernc.org/mathutil v1.7.1 // indirect
 	modernc.org/memory v1.11.0 // indirect

go.sum

@@ -102,49 +102,49 @@ github.com/yuin/goldmark v1.4.13/go.mod h1:6yULJ656Px+3vBD8DxQVa3kxgyrAnzto9xy5t
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20210921155107-089bfa567519/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
-golang.org/x/crypto v0.51.0 h1:IBPXwPfKxY7cWQZ38ZCIRPI50YLeevDLlLnyC5wRGTI=
-golang.org/x/crypto v0.51.0/go.mod h1:8AdwkbraGNABw2kOX6YFPs3WM22XqI4EXEd8g+x7Oc8=
+golang.org/x/crypto v0.53.0 h1:QZ4Muo8THX6CizN2vPPd5fBGHyogrdK9fG4wLPFUsto=
+golang.org/x/crypto v0.53.0/go.mod h1:DNLU434OwVakk9PzuwV8w62mAJpRJL3vsgcfp4Qnsio=
 golang.org/x/mod v0.6.0-dev.0.20220419223038-86c51ed26bb4/go.mod h1:jJ57K6gSWd91VN4djpZkiMVwK6gcyfeH4XE8wZrZaV4=
 golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
-golang.org/x/mod v0.36.0 h1:JJjpVx6myfUsUdAzZuOSTTmRE0PfZeNWzzvKrP7amb4=
-golang.org/x/mod v0.36.0/go.mod h1:moc6ELqsWcOw5Ef3xVprK5ul/MvtVvkIXLziUOICjUQ=
+golang.org/x/mod v0.37.0 h1:vF1DjpVEshcIqoEaauuHebaLk1O1forxjxBaVn884JQ=
+golang.org/x/mod v0.37.0/go.mod h1:m8S8VeM9r4dzDwjrKO0a1sZP3YjeMamRRlD+fmR2Q/0=
 golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c=
 golang.org/x/net v0.6.0/go.mod h1:2Tu9+aMcznHK/AK1HMvgo6xiTLG5rD5rZLDS+rp2Bjs=
-golang.org/x/net v0.54.0 h1:2zJIZAxAHV/OHCDTCOHAYehQzLfSXuf/5SoL/Dv6w/w=
-golang.org/x/net v0.54.0/go.mod h1:Sj4oj8jK6XmHpBZU/zWHw3BV3abl4Kvi+Ut7cQcY+cQ=
+golang.org/x/net v0.56.0 h1:Rw8j/hFzGvJUZwNBXnAtf5sVDVt+65SK2C7IxCxZt5o=
+golang.org/x/net v0.56.0/go.mod h1:D3Ku6r+V6JROoZK144D2XfMHFcMq/0zSfLelVTCFKec=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.1.0/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
-golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sync v0.21.0 h1:HLII4xRRTtCRkxYp4HNFF0Js/Og6q2i++KXbg0gHCwM=
+golang.org/x/sync v0.21.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.44.0 h1:ildZl3J4uzeKP07r2F++Op7E9B29JRUy+a27EibtBTQ=
-golang.org/x/sys v0.44.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/sys v0.46.0 h1:noSf2Fq6F8DBgS+LysIkx7rIExoNHJsxOAtPp4rthXw=
+golang.org/x/sys v0.46.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
 golang.org/x/term v0.5.0/go.mod h1:jMB1sMXY+tzblOD4FWmEbocvup2/aLOaQEp7JmGp78k=
-golang.org/x/term v0.43.0 h1:S4RLU2sB31O/NCl+zFN9Aru9A/Cq2aqKpTZJ6B+DwT4=
-golang.org/x/term v0.43.0/go.mod h1:lrhlHNdQJHO+1qVYiHfFKVuVioJIheAc3fBSMFYEIsk=
+golang.org/x/term v0.44.0 h1:0rLvDRCtNj0gZkyIXhCyOb2OAzEhLVqc4B+hrsBhrmc=
+golang.org/x/term v0.44.0/go.mod h1:7ze4MdzUzLXpSAoFP1H0bOI9aXDqveSvatT5vKcFh2Y=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=
 golang.org/x/text v0.7.0/go.mod h1:mrYo+phRRbMaCq/xk9113O4dZlRixOauAjOtrjsXDZ8=
 golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
-golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
-golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
+golang.org/x/text v0.39.0 h1:UbZz4pLOvn600D6Oh6GGEI6VAmndrEBLv8/6BEXzyus=
+golang.org/x/text v0.39.0/go.mod h1:3UwRclnC2g0TU9x8PZiyfOajCd1zaUNHF9cvqcQZ+ZM=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc=
 golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU=
-golang.org/x/tools v0.45.0 h1:18qN3FAooORvApf5XjCXgsuayZOEtXf6JK18I3+ONa8=
-golang.org/x/tools v0.45.0/go.mod h1:LuUGqqaXcXMEFEruIVJVm5mgDD8vww/z/SR1gQ4uE/0=
+golang.org/x/tools v0.47.0 h1:7Kn5x/d1svx/PzryTsqeoZN4TZwqeH5pGWjefhLi/1Q=
+golang.org/x/tools v0.47.0/go.mod h1:dFHnyTvFWY212G+h7ZY4Vsp/K3U4/7W9TyVaAul8uCA=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=

linux/arch/PKGBUILD

@@ -1,6 +1,6 @@
 # Maintainer: Hein (Warky Devs) <hein@warky.dev>
 pkgname=relspec
-pkgver=1.0.71
+pkgver=1.0.74
 pkgrel=1
 pkgdesc="RelSpec is a comprehensive database relations management tool that reads, transforms, and writes database table specifications across multiple formats and ORMs."
 arch=('x86_64' 'aarch64')

linux/centos/relspec.spec

@@ -1,5 +1,5 @@
 Name:           relspec
-Version:        1.0.71
+Version:        1.0.74
 Release:        1%{?dist}
 Summary:        RelSpec is a comprehensive database relations management tool that reads, transforms, and writes database table specifications across multiple formats and ORMs.
 

pkg/assetloader/loader_test.go

@@ -137,7 +137,10 @@ func TestScanDir_OrdersByPriorityThenSequence(t *testing.T) {
 		t.Fatalf("expected 4 items, got %d", len(items))
 	}
 
-	type ps struct{ p int; s uint }
+	type ps struct {
+		p int
+		s uint
+	}
 	want := []ps{{1, 1}, {1, 2}, {2, 1}, {2, 2}}
 	for i, w := range want {
 		got := ps{items[i].Priority, items[i].Sequence}

pkg/diff/diff.go

@@ -4,6 +4,8 @@ import (
 	"fmt"
 	"reflect"
 	"sort"
+	"strconv"
+	"strings"
 
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
 )
@@ -229,11 +231,13 @@ func compareColumns(source, target map[string]*models.Column) *ColumnDiff {
 
 func compareColumnDetails(source, target *models.Column) map[string]any {
 	changes := make(map[string]any)
+	sourceType, sourceLength, sourceDefault := comparableColumn(source)
+	targetType, targetLength, targetDefault := comparableColumn(target)
 
-	if source.Type != target.Type {
+	if sourceType != targetType {
 		changes["type"] = map[string]string{"source": source.Type, "target": target.Type}
 	}
-	if source.Length != target.Length {
+	if sourceLength != targetLength {
 		changes["length"] = map[string]int{"source": source.Length, "target": target.Length}
 	}
 	if source.Precision != target.Precision {
@@ -245,8 +249,8 @@ func compareColumnDetails(source, target *models.Column) map[string]any {
 	if source.NotNull != target.NotNull {
 		changes["not_null"] = map[string]bool{"source": source.NotNull, "target": target.NotNull}
 	}
-	if !reflect.DeepEqual(source.Default, target.Default) {
-		changes["default"] = map[string]any{"source": source.Default, "target": target.Default}
+	if !reflect.DeepEqual(sourceDefault, targetDefault) {
+		changes["default"] = map[string]any{"source": sourceDefault, "target": targetDefault}
 	}
 	if source.AutoIncrement != target.AutoIncrement {
 		changes["auto_increment"] = map[string]bool{"source": source.AutoIncrement, "target": target.AutoIncrement}
@@ -258,6 +262,28 @@ func compareColumnDetails(source, target *models.Column) map[string]any {
 	return changes
 }
 
+// comparableColumn accepts DBML's compact type/default spelling as well as
+// PostgreSQL's normalized fields (for example varchar(255) vs varchar + 255).
+func comparableColumn(column *models.Column) (normalizedType string, length int, defaultVal any) {
+	typeName := strings.TrimSpace(column.Type)
+	defaultValue := column.Default
+	lower := strings.ToLower(typeName)
+	if i := strings.Index(lower, " default "); i >= 0 {
+		if defaultValue == nil {
+			defaultValue = strings.TrimSpace(typeName[i+len(" default "):])
+		}
+		typeName = strings.TrimSpace(typeName[:i])
+	}
+	length = column.Length
+	if open := strings.LastIndex(typeName, "("); open >= 0 && strings.HasSuffix(typeName, ")") {
+		if parsed, err := strconv.Atoi(strings.TrimSpace(typeName[open+1 : len(typeName)-1])); err == nil && length == 0 {
+			length = parsed
+		}
+		typeName = strings.TrimSpace(typeName[:open])
+	}
+	return strings.ToLower(typeName), length, defaultValue
+}
+
 func compareIndexes(source, target map[string]*models.Index) *IndexDiff {
 	diff := &IndexDiff{
 		Missing:  make([]*models.Index, 0),
@@ -265,12 +291,27 @@ func compareIndexes(source, target map[string]*models.Index) *IndexDiff {
 		Modified: make([]*IndexChange, 0),
 	}
 
-	// Find missing and modified indexes
+	// Match by name first, then by definition. PostgreSQL and DBML can assign
+	// different names to the same index (for example, posts_user_id_title_idx
+	// and uidx_posts_user_id_title), so a name-only comparison reports false
+	// drift after a merge/diff round trip.
+	unmatchedSource := make(map[string]*models.Index, len(source))
+	unmatchedTarget := make(map[string]*models.Index, len(target))
+	for name, index := range source {
+		unmatchedSource[name] = index
+	}
+	for name, index := range target {
+		unmatchedTarget[name] = index
+	}
+
 	for _, name := range sortedKeys(source) {
 		srcIdx := source[name]
-		if tgtIdx, exists := target[name]; !exists {
-			diff.Missing = append(diff.Missing, srcIdx)
-		} else {
+		tgtIdx, exists := target[name]
+		if !exists {
+			continue
+		}
+		delete(unmatchedSource, name)
+		delete(unmatchedTarget, name)
 		if changes := compareIndexDetails(srcIdx, tgtIdx); len(changes) > 0 {
 			diff.Modified = append(diff.Modified, &IndexChange{
 				Name:    name,
@@ -280,19 +321,55 @@ func compareIndexes(source, target map[string]*models.Index) *IndexDiff {
 			})
 		}
 	}
-	}
 
-	// Find extra indexes
-	for _, name := range sortedKeys(target) {
-		tgtIdx := target[name]
-		if _, exists := source[name]; !exists {
-			diff.Extra = append(diff.Extra, tgtIdx)
+	// Pair remaining indexes by their structural identity, independent of the
+	// generated/name field. The sorted iteration makes ambiguous matches
+	// deterministic; duplicate definitions are still represented as separate
+	// indexes by consuming one target at a time.
+	remainingTarget := make(map[string][]*models.Index)
+	for _, name := range sortedKeys(unmatchedTarget) {
+		index := unmatchedTarget[name]
+		key := indexDefinitionKey(index)
+		remainingTarget[key] = append(remainingTarget[key], index)
+	}
+	for _, name := range sortedKeys(unmatchedSource) {
+		srcIdx := unmatchedSource[name]
+		key := indexDefinitionKey(srcIdx)
+		candidates := remainingTarget[key]
+		if len(candidates) == 0 {
+			diff.Missing = append(diff.Missing, srcIdx)
+			continue
+		}
+		tgtIdx := candidates[0]
+		remainingTarget[key] = candidates[1:]
+		if changes := compareIndexDetails(srcIdx, tgtIdx); len(changes) > 0 {
+			diff.Modified = append(diff.Modified, &IndexChange{
+				Name:    srcIdx.Name,
+				Source:  srcIdx,
+				Target:  tgtIdx,
+				Changes: changes,
+			})
 		}
 	}
 
+	for _, key := range sortedKeys(remainingTarget) {
+		diff.Extra = append(diff.Extra, remainingTarget[key]...)
+	}
 	return diff
 }
 
+func indexDefinitionKey(index *models.Index) string {
+	return fmt.Sprintf("%t:%s:%s", index.Unique, strings.Join(index.Columns, ","), strings.Join(index.Include, ","))
+}
+
+func comparableIndexType(indexType string) string {
+	indexType = strings.ToLower(strings.TrimSpace(indexType))
+	if indexType == "" {
+		return "btree"
+	}
+	return indexType
+}
+
 func compareIndexDetails(source, target *models.Index) map[string]any {
 	changes := make(map[string]any)
 
@@ -302,7 +379,7 @@ func compareIndexDetails(source, target *models.Index) map[string]any {
 	if source.Unique != target.Unique {
 		changes["unique"] = map[string]bool{"source": source.Unique, "target": target.Unique}
 	}
-	if source.Type != target.Type {
+	if comparableIndexType(source.Type) != comparableIndexType(target.Type) {
 		changes["type"] = map[string]string{"source": source.Type, "target": target.Type}
 	}
 	if source.Where != target.Where {
@@ -312,7 +389,26 @@ func compareIndexDetails(source, target *models.Index) map[string]any {
 	return changes
 }
 
+// Compare constraints.
+// Primary-key constraints are excluded: a PK is already represented by the
+// column's IsPrimaryKey flag, which compareColumns already compares. The
+// PostgreSQL reader additionally materialises each PK as a primary_key
+// constraint and a unique btree index; the DBML reader keeps PKs as column
+// flags only. Comparing the constraint maps directly would therefore report
+// every PK as an "extra" constraint and the generated index as an "extra"
+// index on a freshly-applied schema. Filtering them here keeps the round
+// trip stable without losing real PK information.
 func compareConstraints(source, target map[string]*models.Constraint) *ConstraintDiff {
+	filteredSource := filterPrimaryKeyConstraints(source)
+	filteredTarget := filterPrimaryKeyConstraints(target)
+	sourceByKey := make(map[string]*models.Constraint, len(filteredSource))
+	targetByKey := make(map[string]*models.Constraint, len(filteredTarget))
+	for _, constraint := range filteredSource {
+		sourceByKey[constraintCompareKey(constraint)] = constraint
+	}
+	for _, constraint := range filteredTarget {
+		targetByKey[constraintCompareKey(constraint)] = constraint
+	}
 	diff := &ConstraintDiff{
 		Missing:  make([]*models.Constraint, 0),
 		Extra:    make([]*models.Constraint, 0),
@@ -320,9 +416,9 @@ func compareConstraints(source, target map[string]*models.Constraint) *Constrain
 	}
 
 	// Find missing and modified constraints
-	for _, name := range sortedKeys(source) {
-		srcCon := source[name]
-		if tgtCon, exists := target[name]; !exists {
+	for _, name := range sortedKeys(sourceByKey) {
+		srcCon := sourceByKey[name]
+		if tgtCon, exists := targetByKey[name]; !exists {
 			diff.Missing = append(diff.Missing, srcCon)
 		} else {
 			if changes := compareConstraintDetails(srcCon, tgtCon); len(changes) > 0 {
@@ -337,9 +433,9 @@ func compareConstraints(source, target map[string]*models.Constraint) *Constrain
 	}
 
 	// Find extra constraints
-	for _, name := range sortedKeys(target) {
-		tgtCon := target[name]
-		if _, exists := source[name]; !exists {
+	for _, name := range sortedKeys(targetByKey) {
+		tgtCon := targetByKey[name]
+		if _, exists := sourceByKey[name]; !exists {
 			diff.Extra = append(diff.Extra, tgtCon)
 		}
 	}
@@ -347,6 +443,29 @@ func compareConstraints(source, target map[string]*models.Constraint) *Constrain
 	return diff
 }
 
+// filterPrimaryKeyConstraints drops primary_key constraints from a single
+// map. Primary keys are compared by the column IsPrimaryKey flag in
+// compareColumns, so comparing the primary_key constraints here only
+// produces duplicate "extra" entries (every PK is extra on the DBML side).
+// Other constraint types are preserved untouched.
+func filterPrimaryKeyConstraints(m map[string]*models.Constraint) map[string]*models.Constraint {
+	out := make(map[string]*models.Constraint, len(m))
+	for name, c := range m {
+		if c.Type == models.PrimaryKeyConstraint {
+			continue
+		}
+		out[name] = c
+	}
+	return out
+}
+
+func constraintCompareKey(constraint *models.Constraint) string {
+	if constraint.Type != models.ForeignKeyConstraint {
+		return constraint.SQLName()
+	}
+	return fmt.Sprintf("fk:%s:%s:%s:%s:%s:%s", strings.ToLower(constraint.Schema), strings.ToLower(constraint.Table), strings.Join(constraint.Columns, ","), strings.ToLower(constraint.ReferencedSchema), strings.ToLower(constraint.ReferencedTable), strings.Join(constraint.ReferencedColumns, ","))
+}
+
 func compareConstraintDetails(source, target *models.Constraint) map[string]any {
 	changes := make(map[string]any)
 
@@ -362,16 +481,23 @@ func compareConstraintDetails(source, target *models.Constraint) map[string]any
 	if !reflect.DeepEqual(source.ReferencedColumns, target.ReferencedColumns) {
 		changes["referenced_columns"] = map[string][]string{"source": source.ReferencedColumns, "target": target.ReferencedColumns}
 	}
-	if source.OnDelete != target.OnDelete {
+	if normalizeConstraintAction(source.OnDelete) != normalizeConstraintAction(target.OnDelete) {
 		changes["on_delete"] = map[string]string{"source": source.OnDelete, "target": target.OnDelete}
 	}
-	if source.OnUpdate != target.OnUpdate {
+	if normalizeConstraintAction(source.OnUpdate) != normalizeConstraintAction(target.OnUpdate) {
 		changes["on_update"] = map[string]string{"source": source.OnUpdate, "target": target.OnUpdate}
 	}
 
 	return changes
 }
 
+func normalizeConstraintAction(action string) string {
+	if strings.EqualFold(strings.TrimSpace(action), "NO ACTION") {
+		return ""
+	}
+	return strings.ToUpper(strings.TrimSpace(action))
+}
+
 func compareRelationships(source, target map[string]*models.Relationship) *RelationshipDiff {
 	diff := &RelationshipDiff{
 		Missing:  make([]*models.Relationship, 0),

pkg/diff/diff_test.go

@@ -301,6 +301,22 @@ func TestCompareIndexes(t *testing.T) {
 				return len(d.Modified) == 1 && d.Modified[0].Name == "idx_name"
 			},
 		},
+		{
+			name: "equivalent indexes with different generated names",
+			source: map[string]*models.Index{
+				"uidx_posts_user_id_title": {
+					Name: "uidx_posts_user_id_title", Columns: []string{"user_id", "title"}, Unique: true,
+				},
+			},
+			target: map[string]*models.Index{
+				"posts_user_id_title_idx": {
+					Name: "posts_user_id_title_idx", Columns: []string{"user_id", "title"}, Unique: true, Type: "btree",
+				},
+			},
+			want: func(d *IndexDiff) bool {
+				return len(d.Missing) == 0 && len(d.Extra) == 0 && len(d.Modified) == 0
+			},
+		},
 	}
 
 	for _, tt := range tests {

pkg/diff/formatters_test.go

@@ -130,7 +130,6 @@ func TestFormatSummary(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			var buf bytes.Buffer
 			err := formatSummary(tt.result, &buf)
-
 			if err != nil {
 				t.Errorf("formatSummary() error = %v", err)
 				return
@@ -159,7 +158,6 @@ func TestFormatJSON(t *testing.T) {
 
 	var buf bytes.Buffer
 	err := formatJSON(result, &buf)
-
 	if err != nil {
 		t.Errorf("formatJSON() error = %v", err)
 		return
@@ -288,7 +286,6 @@ func TestFormatHTML(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			var buf bytes.Buffer
 			err := formatHTML(tt.result, &buf)
-
 			if err != nil {
 				t.Errorf("formatHTML() error = %v", err)
 				return
@@ -334,7 +331,6 @@ func TestFormatSummaryWithColumns(t *testing.T) {
 
 	var buf bytes.Buffer
 	err := formatSummary(result, &buf)
-
 	if err != nil {
 		t.Errorf("formatSummary() error = %v", err)
 		return
@@ -383,7 +379,6 @@ func TestFormatSummaryWithIndexes(t *testing.T) {
 
 	var buf bytes.Buffer
 	err := formatSummary(result, &buf)
-
 	if err != nil {
 		t.Errorf("formatSummary() error = %v", err)
 		return
@@ -425,7 +420,6 @@ func TestFormatSummaryWithConstraints(t *testing.T) {
 
 	var buf bytes.Buffer
 	err := formatSummary(result, &buf)
-
 	if err != nil {
 		t.Errorf("formatSummary() error = %v", err)
 		return
@@ -448,7 +442,6 @@ func TestFormatJSONIndentation(t *testing.T) {
 
 	var buf bytes.Buffer
 	err := formatJSON(result, &buf)
-
 	if err != nil {
 		t.Errorf("formatJSON() error = %v", err)
 		return

pkg/inspector/inspector.go

@@ -168,7 +168,7 @@ func getValidator(functionName string) (validatorFunc, bool) {
 }
 
 // createResult is a helper to create a validation result
-func createResult(ruleName string, passed bool, message string, location string, context map[string]interface{}) ValidationResult {
+func createResult(ruleName string, passed bool, message, location string, context map[string]interface{}) ValidationResult {
 	return ValidationResult{
 		RuleName: ruleName,
 		Message:  message,

pkg/inspector/inspector_test.go

@@ -29,7 +29,6 @@ func TestInspect(t *testing.T) {
 
 	inspector := NewInspector(db, config)
 	report, err := inspector.Inspect()
-
 	if err != nil {
 		t.Fatalf("Inspect() returned error: %v", err)
 	}
@@ -103,7 +102,6 @@ func TestInspectWithDisabledRules(t *testing.T) {
 
 	inspector := NewInspector(db, config)
 	report, err := inspector.Inspect()
-
 	if err != nil {
 		t.Fatalf("Inspect() with disabled rules returned error: %v", err)
 	}
@@ -135,7 +133,6 @@ func TestInspectWithEnforcedRules(t *testing.T) {
 
 	inspector := NewInspector(db, config)
 	report, err := inspector.Inspect()
-
 	if err != nil {
 		t.Fatalf("Inspect() returned error: %v", err)
 	}

pkg/inspector/report.go

@@ -141,7 +141,7 @@ func (f *MarkdownFormatter) formatHeader(text string) string {
 	return f.formatBold("# " + text)
 }
 
-func (f *MarkdownFormatter) formatSubheader(text string, color string) string {
+func (f *MarkdownFormatter) formatSubheader(text, color string) string {
 	header := "### " + text
 	if f.UseColors {
 		return color + colorBold + header + colorReset
@@ -156,7 +156,7 @@ func (f *MarkdownFormatter) formatBold(text string) string {
 	return "**" + text + "**"
 }
 
-func (f *MarkdownFormatter) colorize(text string, color string) string {
+func (f *MarkdownFormatter) colorize(text, color string) string {
 	if f.UseColors {
 		return color + text + colorReset
 	}

pkg/inspector/rules_test.go

@@ -49,7 +49,6 @@ func TestGetDefaultConfig(t *testing.T) {
 func TestLoadConfig_NonExistentFile(t *testing.T) {
 	// Try to load a non-existent file
 	config, err := LoadConfig("/path/to/nonexistent/file.yaml")
-
 	if err != nil {
 		t.Fatalf("LoadConfig() with non-existent file returned error: %v", err)
 	}
@@ -83,7 +82,7 @@ rules:
     message: "Table name too long"
 `
 
-	err := os.WriteFile(configPath, []byte(configContent), 0644)
+	err := os.WriteFile(configPath, []byte(configContent), 0o644)
 	if err != nil {
 		t.Fatalf("Failed to create test config file: %v", err)
 	}
@@ -133,7 +132,7 @@ func TestLoadConfig_InvalidYAML(t *testing.T) {
 
 	invalidContent := `invalid: yaml: content: {[}]`
 
-	err := os.WriteFile(configPath, []byte(invalidContent), 0644)
+	err := os.WriteFile(configPath, []byte(invalidContent), 0o644)
 	if err != nil {
 		t.Fatalf("Failed to create test config file: %v", err)
 	}

pkg/jobs/jobs.go

@@ -0,0 +1,954 @@
+// Package jobs implements RelSpec declarative job files.
+//
+// A job file is a small YAML manifest that names one or more jobs and,
+// for each job, the RelSpec command to run plus its inputs, output and
+// options. It lets users run "relspec job run build-schema" instead of
+// repeating long command lines.
+//
+// The job-file system is deliberately NOT a shell: "command" is a closed
+// enum of vetted RelSpec workflows, every path is resolved relative to the
+// directory holding the job file and may not escape it, and remote database
+// credentials are referenced by environment-variable name only - never
+// embedded in the manifest. All discovery, parsing and validation in this
+// package is side-effect free; nothing here reads input schemas, opens
+// database connections or writes output. Execution lives in the CLI layer
+// and only runs after Validate and the caller's pre-flight checks pass.
+package jobs
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"sort"
+	"strconv"
+	"strings"
+
+	"gopkg.in/yaml.v3"
+)
+
+// CurrentSchemaVersion is the highest job-file schema version this build was
+// written for. MinSchemaVersion is the oldest it still accepts. A file that
+// declares a version in between loads normally; a newer version loads
+// best-effort with a warning (see Load); an older-than-minimum version is a
+// hard error.
+const (
+	CurrentSchemaVersion = 1
+	MinSchemaVersion     = 1
+)
+
+// Built-in logfile rotation policy, used when neither the job nor its file's
+// defaults block sets one.
+const (
+	defaultLogMaxSizeBytes int64 = 5 << 20 // 5 MiB
+	defaultLogKeep               = 3
+)
+
+// Command names are a closed allow-list. Arbitrary strings are rejected.
+const (
+	CommandConvert     = "convert"      // read one or more schema files, optionally merge, write one output
+	CommandMerge       = "merge"        // additive merge of two or more schema files into one output
+	CommandScriptsList = "scripts-list" // deterministically list SQL scripts across one or more directories
+	CommandScriptsExec = "scripts-exec" // execute SQL scripts across one or more directories against a live database
+	CommandTempl       = "templ"        // apply a custom Go text template to one or more schemas
+	CommandSplit       = "split"        // extract selected schemas/tables into a separate output
+	CommandInspect     = "inspect"      // validate one or more schemas against rules and write a report
+	CommandDiff        = "diff"         // compare exactly two schemas and write a differences report
+)
+
+// SupportedCommands lists every accepted command, in help order.
+var SupportedCommands = []string{
+	CommandConvert, CommandMerge, CommandScriptsList, CommandScriptsExec,
+	CommandTempl, CommandSplit, CommandInspect, CommandDiff,
+}
+
+// producerCommands are commands whose output is a schema file that another job
+// may consume via from_job.
+var producerCommands = map[string]bool{
+	CommandConvert: true, CommandMerge: true, CommandSplit: true,
+}
+
+// readerFormats are the file-based input formats a job may declare (path).
+var readerFormats = map[string]bool{
+	"dbml": true, "dctx": true, "drawdb": true, "graphql": true, "json": true,
+	"yaml": true, "gorm": true, "bun": true, "drizzle": true, "prisma": true,
+	"typeorm": true, "sqlite": true,
+}
+
+// inputDBFormats are input formats that can only come from a live connection,
+// referenced by conn_env.
+var inputDBFormats = map[string]bool{"pgsql": true, "mssql": true}
+
+// writerFormats are the output formats a job may declare.
+var writerFormats = map[string]bool{
+	"dbml": true, "dctx": true, "drawdb": true, "graphql": true, "json": true,
+	"yaml": true, "gorm": true, "bun": true, "drizzle": true, "prisma": true,
+	"typeorm": true, "pgsql": true, "mssql": true, "sqlite": true,
+}
+
+// execOutputFormats are output formats for which conn_env (execute against a
+// live database) is supported instead of writing a file.
+var execOutputFormats = map[string]bool{"pgsql": true}
+
+// singleFileFormats are output formats that emit exactly one file (as opposed
+// to a directory of files). Only these are eligible for atomic temp+rename
+// writes and for being consumed by another job via from_job.
+var singleFileFormats = map[string]bool{
+	"json": true, "yaml": true, "dbml": true, "dctx": true, "drawdb": true,
+	"graphql": true, "pgsql": true, "mssql": true, "sqlite": true,
+}
+
+// SingleFileOutputFormat reports whether format writes exactly one file.
+func SingleFileOutputFormat(format string) bool {
+	return singleFileFormats[strings.ToLower(format)]
+}
+
+// diffReportFormats and inspectReportFormats are the report.format values
+// accepted by the diff and inspect commands respectively.
+var (
+	diffReportFormats    = map[string]bool{"summary": true, "json": true, "html": true}
+	inspectReportFormats = map[string]bool{"markdown": true, "json": true}
+)
+
+// File is the on-disk shape of a single job file.
+type File struct {
+	Version  int             `yaml:"version"`
+	Defaults *Defaults       `yaml:"defaults"`
+	Jobs     map[string]*Job `yaml:"jobs"`
+}
+
+// Defaults carries file-wide settings that individual jobs may override.
+type Defaults struct {
+	// LogMaxSize is a human-readable size ("5MB", "512KB", "1GB"). Empty
+	// means "use the built-in default".
+	LogMaxSize string `yaml:"log_max_size"`
+	// LogKeep is how many rotated logfiles to retain. Zero means "use the
+	// built-in default".
+	LogKeep int `yaml:"log_keep"`
+}
+
+// Job is one named job within a job file.
+type Job struct {
+	// Name and SourceFile are populated by Load, not parsed from YAML.
+	Name       string `yaml:"-"`
+	SourceFile string `yaml:"-"`
+	// fileDefaults is the Defaults block of the file that declared this job,
+	// captured by Load. nil when the file had none.
+	fileDefaults *Defaults `yaml:"-"`
+
+	Command         string   `yaml:"command"`
+	Description     string   `yaml:"description"`
+	DependsOn       []string `yaml:"depends_on"`
+	Inputs          []Input  `yaml:"inputs"`
+	ScriptDirs      []string `yaml:"script_dirs"`
+	Template        string   `yaml:"template"`
+	Mode            string   `yaml:"mode"`
+	FilenamePattern string   `yaml:"filename_pattern"`
+	Output          *Output  `yaml:"output"`
+	Rules           string   `yaml:"rules"`
+	Report          *Report  `yaml:"report"`
+	Select          *Select  `yaml:"select"`
+	Options         Options  `yaml:"options"`
+	Logfile         string   `yaml:"logfile"`
+	LogMaxSize      string   `yaml:"log_max_size"`
+	LogKeep         *int     `yaml:"log_keep"`
+}
+
+// Input is one declared input schema.
+type Input struct {
+	Path string `yaml:"path"`
+	// Format is the RelSpec reader format (dbml, json, yaml, pgsql, ...).
+	Format string `yaml:"format"`
+	// ConnEnv is the NAME of an environment variable holding a connection
+	// string, used with database formats. The value is never stored here.
+	ConnEnv string `yaml:"conn_env"`
+	// FromJob names another job in the set whose file output is used as this
+	// input. It implies a dependency on that job. Path/Format/ConnEnv must be
+	// empty when FromJob is set; the format is inherited from the producer.
+	FromJob string `yaml:"from_job"`
+}
+
+// Output is the declared output target.
+type Output struct {
+	Format    string `yaml:"format"`
+	Path      string `yaml:"path"`
+	ConnEnv   string `yaml:"conn_env"`
+	Overwrite bool   `yaml:"overwrite"`
+}
+
+// Report is the output target for the inspect and diff commands.
+type Report struct {
+	// Format is the report format: diff accepts summary|json|html, inspect
+	// accepts markdown|json. Empty means the command's default.
+	Format    string `yaml:"format"`
+	Path      string `yaml:"path"`
+	Overwrite bool   `yaml:"overwrite"`
+}
+
+// Select carries the schema/table selection for the split command.
+type Select struct {
+	Schemas        []string `yaml:"schemas"`
+	Tables         []string `yaml:"tables"`
+	ExcludeSchemas []string `yaml:"exclude_schemas"`
+	ExcludeTables  []string `yaml:"exclude_tables"`
+	DatabaseName   string   `yaml:"database_name"`
+}
+
+// LogPolicy is the resolved logfile rotation policy for a job.
+type LogPolicy struct {
+	MaxSizeBytes int64
+	Keep         int
+}
+
+// ResolvedLogPolicy returns the effective rotation policy: the job's own
+// overrides win, then its file's defaults block, then the built-in default.
+func (j *Job) ResolvedLogPolicy() LogPolicy {
+	p := LogPolicy{MaxSizeBytes: defaultLogMaxSizeBytes, Keep: defaultLogKeep}
+
+	if j.fileDefaults != nil {
+		if n, err := parseHumanSize(j.fileDefaults.LogMaxSize); err == nil && n > 0 {
+			p.MaxSizeBytes = n
+		}
+		if j.fileDefaults.LogKeep > 0 {
+			p.Keep = j.fileDefaults.LogKeep
+		}
+	}
+	if n, err := parseHumanSize(j.LogMaxSize); err == nil && n > 0 {
+		p.MaxSizeBytes = n
+	}
+	if j.LogKeep != nil && *j.LogKeep >= 0 {
+		p.Keep = *j.LogKeep
+	}
+	return p
+}
+
+// effectiveDeps returns the union of explicit depends_on entries and the jobs
+// referenced by from_job inputs, deduplicated in stable order.
+func (j *Job) effectiveDeps() []string {
+	seen := map[string]bool{}
+	var deps []string
+	add := func(name string) {
+		if name == "" || name == j.Name || seen[name] {
+			return
+		}
+		seen[name] = true
+		deps = append(deps, name)
+	}
+	for _, d := range j.DependsOn {
+		add(d)
+	}
+	for _, in := range j.Inputs {
+		add(in.FromJob)
+	}
+	return deps
+}
+
+// parseHumanSize parses a byte size such as "5MB", "512 KB", "1gb" or a bare
+// byte count. An empty string returns (0, nil) so callers can fall back.
+func parseHumanSize(s string) (int64, error) {
+	s = strings.TrimSpace(s)
+	if s == "" {
+		return 0, nil
+	}
+	upper := strings.ToUpper(s)
+	mult := int64(1)
+	// Check multi-character suffixes before the bare "B".
+	for _, u := range []struct {
+		suffix string
+		m      int64
+	}{
+		{"KB", 1 << 10}, {"MB", 1 << 20}, {"GB", 1 << 30}, {"B", 1},
+	} {
+		if strings.HasSuffix(upper, u.suffix) {
+			mult = u.m
+			upper = strings.TrimSpace(strings.TrimSuffix(upper, u.suffix))
+			break
+		}
+	}
+	n, err := strconv.ParseFloat(upper, 64)
+	if err != nil {
+		return 0, fmt.Errorf("invalid size %q", s)
+	}
+	if n < 0 {
+		return 0, fmt.Errorf("negative size %q", s)
+	}
+	return int64(n * float64(mult)), nil
+}
+
+// Options carries the subset of command flags a job file may set.
+type Options struct {
+	FlattenSchema   bool   `yaml:"flatten_schema"`
+	Schema          string `yaml:"schema"`
+	Package         string `yaml:"package"`
+	ContinueOnError bool   `yaml:"continue_on_error"`
+	SkipRelations   bool   `yaml:"skip_relations"`
+	SkipEnums       bool   `yaml:"skip_enums"`
+	SkipViews       bool   `yaml:"skip_views"`
+	SkipDomains     bool   `yaml:"skip_domains"`
+	SkipSequences   bool   `yaml:"skip_sequences"`
+}
+
+// Dir returns the directory that a job's relative paths resolve against:
+// the directory containing the job file that declared it.
+func (j *Job) Dir() string { return filepath.Dir(j.SourceFile) }
+
+// Set is the merged view of all discovered/selected job files.
+type Set struct {
+	// Files is the sorted list of job files that contributed jobs.
+	Files []string
+	// Jobs is keyed by job name.
+	Jobs map[string]*Job
+	// Warnings holds non-fatal load-time messages (e.g. a newer-than-known
+	// schema version). Callers should surface these to the user.
+	Warnings []string
+}
+
+// Names returns all job names in deterministic (sorted) order.
+func (s *Set) Names() []string {
+	names := make([]string, 0, len(s.Jobs))
+	for n := range s.Jobs {
+		names = append(names, n)
+	}
+	sort.Strings(names)
+	return names
+}
+
+// Discover returns the job files in dir in deterministic order. The default
+// file "relspec.yml"/"relspec.yaml" sorts first, followed by named files
+// "relspec.<name>.yml"/"relspec.<name>.yaml" in lexical order.
+func Discover(dir string) ([]string, error) {
+	if dir == "" {
+		dir = "."
+	}
+	entries, err := os.ReadDir(dir)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read directory %q: %w", dir, err)
+	}
+	var defaults, named []string
+	for _, e := range entries {
+		if e.IsDir() {
+			continue
+		}
+		name := e.Name()
+		if !isJobFileName(name) {
+			continue
+		}
+		full := filepath.Join(dir, name)
+		if name == "relspec.yml" || name == "relspec.yaml" {
+			defaults = append(defaults, full)
+		} else {
+			named = append(named, full)
+		}
+	}
+	sort.Strings(defaults)
+	sort.Strings(named)
+	return append(defaults, named...), nil
+}
+
+func isJobFileName(name string) bool {
+	for _, ext := range []string{".yml", ".yaml"} {
+		if name == "relspec"+ext {
+			return true
+		}
+		if strings.HasPrefix(name, "relspec.") && strings.HasSuffix(name, ext) {
+			return true
+		}
+	}
+	return false
+}
+
+// Load parses every path, rejects unknown fields and unsupported versions,
+// and merges all jobs into one Set. A job name defined by more than one file
+// is a hard error. Load performs structural checks only; call Validate for
+// full semantic validation.
+func Load(paths []string) (*Set, error) {
+	if len(paths) == 0 {
+		return nil, fmt.Errorf("no job files found (looked for relspec.yml / relspec.<name>.yml)")
+	}
+	set := &Set{Jobs: map[string]*Job{}}
+	origin := map[string]string{} // job name -> first file that defined it
+
+	for _, path := range paths {
+		data, err := os.ReadFile(path)
+		if err != nil {
+			return nil, fmt.Errorf("failed to read job file %q: %w", path, err)
+		}
+
+		// Peek at the version first so a newer file can be parsed leniently
+		// (unknown fields ignored) instead of failing outright.
+		var probe struct {
+			Version int `yaml:"version"`
+		}
+		if err := yaml.Unmarshal(data, &probe); err != nil {
+			return nil, fmt.Errorf("invalid job file %q: %w", path, err)
+		}
+		version := probe.Version
+		if version == 0 {
+			version = CurrentSchemaVersion
+		}
+		if version < MinSchemaVersion {
+			return nil, fmt.Errorf("job file %q: unsupported version %d (this build accepts %d or newer)", path, version, MinSchemaVersion)
+		}
+		strict := version <= CurrentSchemaVersion
+		if !strict {
+			set.Warnings = append(set.Warnings, fmt.Sprintf(
+				"job file %q declares version %d, newer than this build understands (%d); loading best-effort and ignoring unknown fields",
+				path, version, CurrentSchemaVersion))
+		}
+
+		dec := yaml.NewDecoder(strings.NewReader(string(data)))
+		dec.KnownFields(strict)
+		var f File
+		if err := dec.Decode(&f); err != nil {
+			return nil, fmt.Errorf("invalid job file %q: %w", path, err)
+		}
+		if len(f.Jobs) == 0 {
+			return nil, fmt.Errorf("job file %q: no jobs defined", path)
+		}
+		for name, job := range f.Jobs {
+			if job == nil {
+				return nil, fmt.Errorf("job file %q: job %q is empty", path, name)
+			}
+			if prev, dup := origin[name]; dup {
+				return nil, fmt.Errorf("duplicate job %q defined in both %q and %q", name, prev, path)
+			}
+			job.Name = name
+			job.SourceFile = path
+			job.fileDefaults = f.Defaults
+			origin[name] = path
+			set.Jobs[name] = job
+		}
+		set.Files = append(set.Files, path)
+	}
+	return set, nil
+}
+
+// Validate runs full semantic validation over the whole set and returns a
+// single error describing every problem found. It never touches the
+// filesystem beyond what Load already read; existence of input files and
+// environment variables is checked by the caller immediately before
+// execution.
+func (s *Set) Validate() error {
+	var errs []string
+	for _, name := range s.Names() {
+		for _, msg := range s.Jobs[name].validate() {
+			errs = append(errs, fmt.Sprintf("job %q: %s", name, msg))
+		}
+	}
+	// Dependency references + cycles + from_job wiring.
+	for _, name := range s.Names() {
+		j := s.Jobs[name]
+		for _, dep := range j.DependsOn {
+			if _, ok := s.Jobs[dep]; !ok {
+				errs = append(errs, fmt.Sprintf("job %q: depends_on unknown job %q", name, dep))
+			}
+		}
+		for i, in := range j.Inputs {
+			if in.FromJob == "" {
+				continue
+			}
+			producer, ok := s.Jobs[in.FromJob]
+			if !ok {
+				errs = append(errs, fmt.Sprintf("job %q: input[%d] from_job references unknown job %q", name, i, in.FromJob))
+				continue
+			}
+			if !producerCommands[producer.Command] || producer.Output == nil ||
+				producer.Output.Path == "" || !SingleFileOutputFormat(producer.Output.Format) {
+				errs = append(errs, fmt.Sprintf(
+					"job %q: input[%d] from_job %q must name a convert/merge/split job that writes a single-file output",
+					name, i, in.FromJob))
+			}
+		}
+	}
+	if cycle := s.findCycle(); cycle != "" {
+		errs = append(errs, fmt.Sprintf("dependency cycle detected: %s", cycle))
+	}
+	if len(errs) > 0 {
+		sort.Strings(errs)
+		return fmt.Errorf("job file validation failed:\n  - %s", strings.Join(errs, "\n  - "))
+	}
+	return nil
+}
+
+func (j *Job) validate() []string {
+	var e []string
+
+	switch j.Command {
+	case CommandConvert, CommandMerge, CommandScriptsList, CommandScriptsExec,
+		CommandTempl, CommandSplit, CommandInspect, CommandDiff:
+	case "":
+		e = append(e, "missing command")
+		return e
+	default:
+		e = append(e, fmt.Sprintf("unsupported command %q (supported: %s)", j.Command, strings.Join(SupportedCommands, ", ")))
+		return e
+	}
+
+	// Path safety for every declared path.
+	checkPath := func(label, p string) {
+		if p == "" {
+			return
+		}
+		if err := checkRelPath(p); err != nil {
+			e = append(e, fmt.Sprintf("%s %q: %v", label, p, err))
+		}
+	}
+	checkPath("logfile", j.Logfile)
+	checkPath("template", j.Template)
+	checkPath("rules", j.Rules)
+	for _, in := range j.Inputs {
+		checkPath("input path", in.Path)
+	}
+	for _, d := range j.ScriptDirs {
+		checkPath("script_dir", d)
+	}
+	if j.Output != nil {
+		checkPath("output path", j.Output.Path)
+	}
+	if j.Report != nil {
+		checkPath("report path", j.Report.Path)
+	}
+
+	if _, err := parseHumanSize(j.LogMaxSize); err != nil {
+		e = append(e, fmt.Sprintf("log_max_size: %v", err))
+	}
+
+	switch j.Command {
+	case CommandConvert, CommandMerge:
+		minInputs := 1
+		if j.Command == CommandMerge {
+			minInputs = 2
+		}
+		if len(j.Inputs) < minInputs {
+			e = append(e, fmt.Sprintf("command %q requires at least %d input(s)", j.Command, minInputs))
+		}
+		for i, in := range j.Inputs {
+			e = append(e, validateInput(i, in)...)
+		}
+		if len(j.ScriptDirs) > 0 {
+			e = append(e, fmt.Sprintf("script_dirs is not valid for command %q", j.Command))
+		}
+		if j.Output == nil {
+			e = append(e, "missing output")
+		} else {
+			e = append(e, validateOutput(*j.Output)...)
+		}
+	case CommandScriptsList:
+		if len(j.ScriptDirs) == 0 {
+			e = append(e, "command \"scripts-list\" requires at least one script_dir")
+		}
+		if len(j.Inputs) > 0 {
+			e = append(e, "inputs is not valid for command \"scripts-list\"")
+		}
+		if j.Output != nil {
+			e = append(e, "output is not valid for command \"scripts-list\"")
+		}
+	case CommandTempl:
+		if len(j.Inputs) < 1 {
+			e = append(e, "command \"templ\" requires at least 1 input")
+		}
+		for i, in := range j.Inputs {
+			e = append(e, validateTemplInput(i, in)...)
+		}
+		if j.Template == "" {
+			e = append(e, "command \"templ\" requires template")
+		}
+		mode := strings.ToLower(j.Mode)
+		if mode == "" {
+			mode = "database"
+		}
+		switch mode {
+		case "database", "schema", "script", "table":
+		default:
+			e = append(e, fmt.Sprintf("command \"templ\" has unsupported mode %q (supported: database, schema, script, table)", j.Mode))
+		}
+		if len(j.ScriptDirs) > 0 {
+			e = append(e, "script_dirs is not valid for command \"templ\"")
+		}
+		if j.Output != nil && j.Output.ConnEnv != "" {
+			e = append(e, "command \"templ\" does not support database output")
+		}
+		if j.Output != nil && j.Output.Format != "" {
+			e = append(e, "output.format is not valid for command \"templ\"")
+		}
+	case CommandSplit:
+		if len(j.Inputs) < 1 {
+			e = append(e, "command \"split\" requires at least 1 input")
+		}
+		for i, in := range j.Inputs {
+			e = append(e, validateInput(i, in)...)
+		}
+		if len(j.ScriptDirs) > 0 {
+			e = append(e, "script_dirs is not valid for command \"split\"")
+		}
+		if j.Report != nil {
+			e = append(e, "report is not valid for command \"split\" (use output)")
+		}
+		if j.Output == nil {
+			e = append(e, "missing output")
+		} else {
+			if j.Output.ConnEnv != "" {
+				e = append(e, "command \"split\" writes a file; output.conn_env is not supported")
+			}
+			e = append(e, validateOutput(*j.Output)...)
+		}
+	case CommandInspect:
+		if len(j.Inputs) < 1 {
+			e = append(e, "command \"inspect\" requires at least 1 input")
+		}
+		for i, in := range j.Inputs {
+			e = append(e, validateInput(i, in)...)
+		}
+		if len(j.ScriptDirs) > 0 {
+			e = append(e, "script_dirs is not valid for command \"inspect\"")
+		}
+		if j.Output != nil {
+			e = append(e, "output is not valid for command \"inspect\" (use report)")
+		}
+		e = append(e, validateReport(j.Report, "inspect", inspectReportFormats, "markdown")...)
+	case CommandDiff:
+		if len(j.Inputs) != 2 {
+			e = append(e, "command \"diff\" requires exactly 2 inputs (source, target)")
+		}
+		for i, in := range j.Inputs {
+			e = append(e, validateInput(i, in)...)
+		}
+		if len(j.ScriptDirs) > 0 {
+			e = append(e, "script_dirs is not valid for command \"diff\"")
+		}
+		if j.Output != nil {
+			e = append(e, "output is not valid for command \"diff\" (use report)")
+		}
+		e = append(e, validateReport(j.Report, "diff", diffReportFormats, "summary")...)
+	case CommandScriptsExec:
+		if len(j.ScriptDirs) == 0 {
+			e = append(e, "command \"scripts-exec\" requires at least one script_dir")
+		}
+		if len(j.Inputs) > 0 {
+			e = append(e, "inputs is not valid for command \"scripts-exec\"")
+		}
+		if j.Report != nil {
+			e = append(e, "report is not valid for command \"scripts-exec\"")
+		}
+		if j.Output == nil || j.Output.ConnEnv == "" {
+			e = append(e, "command \"scripts-exec\" requires output.conn_env (an environment variable name holding a connection string)")
+		} else {
+			if j.Output.Path != "" {
+				e = append(e, "command \"scripts-exec\" executes against a database; output.path is not supported")
+			}
+			f := strings.ToLower(j.Output.Format)
+			if f != "" && f != "pgsql" {
+				e = append(e, fmt.Sprintf("command \"scripts-exec\" only supports pgsql databases (got %q)", j.Output.Format))
+			}
+			if looksLikeSecret(j.Output.ConnEnv) {
+				e = append(e, "output: conn_env must be an environment variable name, not a connection string")
+			}
+		}
+	}
+	return e
+}
+
+// validateReport checks a Report block for the inspect/diff commands.
+func validateReport(r *Report, cmd string, allowed map[string]bool, defFmt string) []string {
+	if r == nil {
+		return []string{fmt.Sprintf("command %q requires a report block", cmd)}
+	}
+	var e []string
+	f := strings.ToLower(r.Format)
+	if f == "" {
+		f = defFmt
+	}
+	if !allowed[f] {
+		names := make([]string, 0, len(allowed))
+		for k := range allowed {
+			names = append(names, k)
+		}
+		sort.Strings(names)
+		e = append(e, fmt.Sprintf("command %q report.format %q is not supported (use: %s)", cmd, r.Format, strings.Join(names, ", ")))
+	}
+	// A diff summary may be written to the log; everything else needs a path.
+	summaryToLog := cmd == "diff" && f == "summary"
+	if r.Path == "" && !summaryToLog {
+		e = append(e, fmt.Sprintf("command %q requires report.path", cmd))
+	}
+	return e
+}
+
+func validateTemplInput(i int, in Input) []string {
+	if in.FromJob != "" {
+		return fromJobInputShape(i, in)
+	}
+	var e []string
+	if in.Format == "" {
+		return []string{fmt.Sprintf("input[%d]: missing format", i)}
+	}
+	f := strings.ToLower(in.Format)
+	if f == "pgsql" {
+		if in.ConnEnv == "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q requires conn_env (an environment variable name)", i, in.Format))
+		}
+		if in.Path != "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q takes conn_env, not path", i, in.Format))
+		}
+	} else if readerFormats[f] {
+		if in.Path == "" {
+			e = append(e, fmt.Sprintf("input[%d]: missing path", i))
+		}
+		if in.ConnEnv != "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q does not use conn_env", i, in.Format))
+		}
+	} else {
+		e = append(e, fmt.Sprintf("input[%d]: unsupported templ input format %q", i, in.Format))
+	}
+	if looksLikeSecret(in.ConnEnv) {
+		e = append(e, fmt.Sprintf("input[%d]: conn_env must be an environment variable name, not a connection string", i))
+	}
+	return e
+}
+
+// fromJobInputShape checks the structural rules for an input that pulls its
+// schema from another job's output. The referenced job's existence and kind
+// are checked in Set.Validate, which can see the whole set.
+func fromJobInputShape(i int, in Input) []string {
+	var e []string
+	if in.Path != "" {
+		e = append(e, fmt.Sprintf("input[%d]: from_job takes no path", i))
+	}
+	if in.Format != "" {
+		e = append(e, fmt.Sprintf("input[%d]: from_job inherits the producer's format; drop format", i))
+	}
+	if in.ConnEnv != "" {
+		e = append(e, fmt.Sprintf("input[%d]: from_job takes no conn_env", i))
+	}
+	return e
+}
+
+func validateInput(i int, in Input) []string {
+	if in.FromJob != "" {
+		return fromJobInputShape(i, in)
+	}
+	var e []string
+	if in.Format == "" {
+		e = append(e, fmt.Sprintf("input[%d]: missing format", i))
+		return e
+	}
+	f := strings.ToLower(in.Format)
+	switch {
+	case inputDBFormats[f]:
+		if in.ConnEnv == "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q requires conn_env (an environment variable name)", i, in.Format))
+		}
+		if in.Path != "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q takes conn_env, not path", i, in.Format))
+		}
+	case readerFormats[f]:
+		if in.Path == "" {
+			e = append(e, fmt.Sprintf("input[%d]: missing path", i))
+		}
+		if in.ConnEnv != "" {
+			e = append(e, fmt.Sprintf("input[%d]: format %q does not use conn_env", i, in.Format))
+		}
+	default:
+		e = append(e, fmt.Sprintf("input[%d]: unsupported input format %q", i, in.Format))
+	}
+	if looksLikeSecret(in.ConnEnv) {
+		e = append(e, fmt.Sprintf("input[%d]: conn_env must be an environment variable name, not a connection string", i))
+	}
+	return e
+}
+
+func validateOutput(o Output) []string {
+	var e []string
+	if o.Format == "" {
+		e = append(e, "output: missing format")
+		return e
+	}
+	f := strings.ToLower(o.Format)
+	if !writerFormats[f] {
+		e = append(e, fmt.Sprintf("output: unsupported output format %q", o.Format))
+		return e
+	}
+	if o.ConnEnv != "" {
+		if !execOutputFormats[f] {
+			e = append(e, fmt.Sprintf("output: conn_env (live database execution) is not supported for format %q", o.Format))
+		}
+		if o.Path != "" {
+			e = append(e, "output: set either path or conn_env, not both")
+		}
+	} else if o.Path == "" {
+		e = append(e, "output: missing path")
+	}
+	if looksLikeSecret(o.ConnEnv) {
+		e = append(e, "output: conn_env must be an environment variable name, not a connection string")
+	}
+	return e
+}
+
+// looksLikeSecret reports whether s looks like a connection string rather
+// than a bare environment-variable name.
+func looksLikeSecret(s string) bool {
+	if s == "" {
+		return false
+	}
+	return strings.ContainsAny(s, ":/@ =") || strings.Contains(s, "//")
+}
+
+// checkRelPath rejects absolute paths and any path that escapes its root.
+func checkRelPath(p string) error {
+	if p == "" {
+		return fmt.Errorf("empty path")
+	}
+	if filepath.IsAbs(p) {
+		return fmt.Errorf("absolute paths are not allowed; use a path relative to the job file")
+	}
+	if strings.HasPrefix(p, "~") {
+		return fmt.Errorf("home-relative paths are not allowed")
+	}
+	clean := filepath.ToSlash(filepath.Clean(p))
+	if clean == ".." || strings.HasPrefix(clean, "../") {
+		return fmt.Errorf("path escapes the job file directory")
+	}
+	return nil
+}
+
+// SafeJoin resolves rel against root and guarantees the result stays inside
+// root. It is the single choke point for turning a manifest path into a
+// filesystem path.
+func SafeJoin(root, rel string) (string, error) {
+	if err := checkRelPath(rel); err != nil {
+		return "", err
+	}
+	absRoot, err := filepath.Abs(root)
+	if err != nil {
+		return "", err
+	}
+	joined := filepath.Join(absRoot, rel)
+	rp, err := filepath.Rel(absRoot, joined)
+	if err != nil {
+		return "", err
+	}
+	if rp == ".." || strings.HasPrefix(rp, ".."+string(filepath.Separator)) {
+		return "", fmt.Errorf("path %q escapes the job file directory", rel)
+	}
+	// Symlink hardening: resolve symlinks on the root and on the deepest
+	// existing ancestor of the target, and require the target to still live
+	// inside the resolved root. This catches a symlink inside the job-file
+	// directory that points outside it.
+	realRoot, err := filepath.EvalSymlinks(absRoot)
+	if err != nil {
+		return "", fmt.Errorf("cannot resolve job file directory: %w", err)
+	}
+	realAnc, err := filepath.EvalSymlinks(deepestExistingAncestor(joined))
+	if err != nil {
+		return "", fmt.Errorf("cannot resolve path %q: %w", rel, err)
+	}
+	if realAnc != realRoot {
+		if r, err := filepath.Rel(realRoot, realAnc); err != nil ||
+			r == ".." || strings.HasPrefix(r, ".."+string(filepath.Separator)) {
+			return "", fmt.Errorf("path %q resolves outside the job file directory via a symlink", rel)
+		}
+	}
+	return joined, nil
+}
+
+// deepestExistingAncestor returns p itself if it exists, otherwise the nearest
+// existing parent directory (falling back to the filesystem root).
+func deepestExistingAncestor(p string) string {
+	for {
+		if _, err := os.Lstat(p); err == nil {
+			return p
+		}
+		parent := filepath.Dir(p)
+		if parent == p {
+			return p
+		}
+		p = parent
+	}
+}
+
+// Plan returns the jobs to execute for name in dependency order. When
+// includeDeps is false only the named job is returned (its declared
+// dependencies are still validated to exist and be acyclic by Validate).
+func (s *Set) Plan(name string, includeDeps bool) ([]*Job, error) {
+	root, ok := s.Jobs[name]
+	if !ok {
+		return nil, fmt.Errorf("unknown job %q (known: %s)", name, strings.Join(s.Names(), ", "))
+	}
+	if !includeDeps {
+		return []*Job{root}, nil
+	}
+	var order []*Job
+	visited := map[string]bool{}
+	inProgress := map[string]bool{}
+	var visit func(n string) error
+	visit = func(n string) error {
+		if visited[n] {
+			return nil
+		}
+		if inProgress[n] {
+			return fmt.Errorf("dependency cycle at job %q", n)
+		}
+		inProgress[n] = true
+		j := s.Jobs[n]
+		deps := j.effectiveDeps()
+		sort.Strings(deps)
+		for _, d := range deps {
+			if _, ok := s.Jobs[d]; !ok {
+				return fmt.Errorf("job %q depends on unknown job %q", n, d)
+			}
+			if err := visit(d); err != nil {
+				return err
+			}
+		}
+		inProgress[n] = false
+		visited[n] = true
+		order = append(order, j)
+		return nil
+	}
+	if err := visit(name); err != nil {
+		return nil, err
+	}
+	return order, nil
+}
+
+// findCycle returns a human-readable cycle path, or "" if the graph is acyclic.
+func (s *Set) findCycle() string {
+	color := map[string]int{} // 0 unvisited, 1 in progress, 2 done
+	var stack []string
+	var dfs func(n string) []string
+	dfs = func(n string) []string {
+		color[n] = 1
+		stack = append(stack, n)
+		deps := s.Jobs[n].effectiveDeps()
+		sort.Strings(deps)
+		for _, d := range deps {
+			if _, ok := s.Jobs[d]; !ok {
+				continue
+			}
+			switch color[d] {
+			case 0:
+				if c := dfs(d); c != nil {
+					return c
+				}
+			case 1:
+				// Found a back edge; build the cycle slice.
+				for i, x := range stack {
+					if x == d {
+						return append(append([]string(nil), stack[i:]...), d)
+					}
+				}
+				return []string{d, d}
+			}
+		}
+		stack = stack[:len(stack)-1]
+		color[n] = 2
+		return nil
+	}
+	for _, n := range s.Names() {
+		if color[n] == 0 {
+			if c := dfs(n); c != nil {
+				return strings.Join(c, " -> ")
+			}
+		}
+	}
+	return ""
+}

pkg/jobs/jobs_test.go

@@ -0,0 +1,449 @@
+package jobs
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func write(t *testing.T, path, content string) {
+	t.Helper()
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(path, []byte(content), 0o644); err != nil {
+		t.Fatal(err)
+	}
+}
+
+func TestDiscoverDeterministicOrder(t *testing.T) {
+	dir := t.TempDir()
+	for _, n := range []string{
+		"relspec.yml", "relspec.zeta.yml", "relspec.alpha.yaml",
+		"relspec.beta.yml", "notes.yml", "relspec.txt",
+	} {
+		write(t, filepath.Join(dir, n), "version: 1\njobs: {}\n")
+	}
+	got, err := Discover(dir)
+	if err != nil {
+		t.Fatal(err)
+	}
+	var bases []string
+	for _, p := range got {
+		bases = append(bases, filepath.Base(p))
+	}
+	want := []string{"relspec.yml", "relspec.alpha.yaml", "relspec.beta.yml", "relspec.zeta.yml"}
+	if strings.Join(bases, ",") != strings.Join(want, ",") {
+		t.Fatalf("discover order = %v, want %v", bases, want)
+	}
+
+	// Second call must return the identical order.
+	got2, _ := Discover(dir)
+	for i := range got {
+		if got[i] != got2[i] {
+			t.Fatalf("discover not deterministic: %v vs %v", got, got2)
+		}
+	}
+}
+
+func TestLoadRejectsUnknownFields(t *testing.T) {
+	dir := t.TempDir()
+	p := filepath.Join(dir, "relspec.yml")
+	write(t, p, "version: 1\njobs:\n  a:\n    command: convert\n    bogus: true\n")
+	if _, err := Load([]string{p}); err == nil {
+		t.Fatal("expected error for unknown field")
+	}
+}
+
+func TestLoadWarnsOnNewerVersion(t *testing.T) {
+	dir := t.TempDir()
+	p := filepath.Join(dir, "relspec.yml")
+	// A newer version loads best-effort with a warning, and unknown fields
+	// from the newer schema are ignored rather than rejected.
+	write(t, p, "version: 99\njobs:\n  a:\n    command: convert\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    output:\n      format: json\n      path: out.json\n"+
+		"    future_field: whatever\n")
+	set, err := Load([]string{p})
+	if err != nil {
+		t.Fatalf("newer version should load, got %v", err)
+	}
+	if len(set.Warnings) == 0 {
+		t.Fatal("expected a warning about the newer version")
+	}
+	if err := set.Validate(); err != nil {
+		t.Fatalf("validate: %v", err)
+	}
+}
+
+func TestLoadAcceptsOmittedVersion(t *testing.T) {
+	dir := t.TempDir()
+	p := filepath.Join(dir, "relspec.yml")
+	write(t, p, "jobs:\n  a:\n    command: convert\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    output:\n      format: json\n      path: out.json\n")
+	set, err := Load([]string{p})
+	if err != nil {
+		t.Fatalf("omitted version should load, got %v", err)
+	}
+	if len(set.Warnings) != 0 {
+		t.Fatalf("omitted version should not warn, got %v", set.Warnings)
+	}
+}
+
+func TestLoadStillRejectsUnknownFieldsAtCurrentVersion(t *testing.T) {
+	dir := t.TempDir()
+	p := filepath.Join(dir, "relspec.yml")
+	write(t, p, "version: 1\njobs:\n  a:\n    command: convert\n    bogus: true\n")
+	if _, err := Load([]string{p}); err == nil {
+		t.Fatal("expected unknown-field rejection at the current version")
+	}
+}
+
+func TestParseHumanSize(t *testing.T) {
+	cases := []struct {
+		in   string
+		want int64
+		bad  bool
+	}{
+		{"", 0, false},
+		{"512", 512, false},
+		{"512B", 512, false},
+		{"1KB", 1 << 10, false},
+		{"5MB", 5 << 20, false},
+		{"1gb", 1 << 30, false},
+		{" 2 MB ", 2 << 20, false},
+		{"nonsense", 0, true},
+		{"-1MB", 0, true},
+	}
+	for _, c := range cases {
+		got, err := parseHumanSize(c.in)
+		if c.bad {
+			if err == nil {
+				t.Errorf("parseHumanSize(%q): expected error", c.in)
+			}
+			continue
+		}
+		if err != nil {
+			t.Errorf("parseHumanSize(%q): %v", c.in, err)
+			continue
+		}
+		if got != c.want {
+			t.Errorf("parseHumanSize(%q) = %d, want %d", c.in, got, c.want)
+		}
+	}
+}
+
+func TestLoadRejectsDuplicateJobAcrossFiles(t *testing.T) {
+	dir := t.TempDir()
+	a := filepath.Join(dir, "relspec.yml")
+	b := filepath.Join(dir, "relspec.extra.yml")
+	write(t, a, jobFileConvert("build"))
+	write(t, b, jobFileConvert("build"))
+	_, err := Load([]string{a, b})
+	if err == nil || !strings.Contains(err.Error(), "duplicate job") {
+		t.Fatalf("expected duplicate job error, got %v", err)
+	}
+}
+
+func jobFileConvert(name string) string {
+	return "version: 1\njobs:\n  " + name + ":\n    command: convert\n" +
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n" +
+		"    output:\n      format: json\n      path: out.json\n"
+}
+
+func loadOne(t *testing.T, content string) *Set {
+	t.Helper()
+	dir := t.TempDir()
+	p := filepath.Join(dir, "relspec.yml")
+	write(t, p, content)
+	set, err := Load([]string{p})
+	if err != nil {
+		t.Fatalf("load: %v", err)
+	}
+	return set
+}
+
+func TestValidateUnknownCommand(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: rm-rf\n")
+	err := set.Validate()
+	if err == nil || !strings.Contains(err.Error(), "unsupported command") {
+		t.Fatalf("want unsupported command, got %v", err)
+	}
+}
+
+func TestValidateShellStringCommandRejected(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: \"bash -c 'echo hi'\"\n")
+	if err := set.Validate(); err == nil {
+		t.Fatal("expected arbitrary shell command to be rejected")
+	}
+}
+
+func TestValidateMissingInputs(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n    output:\n      format: json\n      path: o.json\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "at least 1 input") {
+		t.Fatalf("want missing input error, got %v", err)
+	}
+}
+
+func TestValidateUnknownFormat(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n"+
+		"    inputs:\n      - path: a.xyz\n        format: xyz\n"+
+		"    output:\n      format: json\n      path: o.json\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "unsupported input format") {
+		t.Fatalf("want unsupported input format, got %v", err)
+	}
+}
+
+func TestValidatePathTraversalRejected(t *testing.T) {
+	cases := []string{"../secret.dbml", "/etc/passwd", "~/x.dbml", "a/../../b.dbml"}
+	for _, bad := range cases {
+		set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n"+
+			"    inputs:\n      - path: \""+bad+"\"\n        format: dbml\n"+
+			"    output:\n      format: json\n      path: o.json\n")
+		if err := set.Validate(); err == nil {
+			t.Fatalf("path %q: expected rejection", bad)
+		}
+	}
+}
+
+func TestValidateOutputTraversalRejected(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    output:\n      format: json\n      path: ../../evil.json\n")
+	if err := set.Validate(); err == nil {
+		t.Fatal("expected output path traversal rejection")
+	}
+}
+
+func TestValidateConnEnvMustBeName(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n"+
+		"    inputs:\n      - format: pgsql\n        conn_env: \"postgres://u:p@h/db\"\n"+
+		"    output:\n      format: json\n      path: o.json\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "environment variable name") {
+		t.Fatalf("want conn_env name error, got %v", err)
+	}
+}
+
+func TestValidateDependsOnUnknown(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: convert\n    depends_on: [nope]\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    output:\n      format: json\n      path: o.json\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "unknown job") {
+		t.Fatalf("want unknown dependency error, got %v", err)
+	}
+}
+
+func TestValidateDependencyCycle(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		jobBlock("a", "b") + jobBlock("b", "c") + jobBlock("c", "a")
+	set := loadOne(t, content)
+	err := set.Validate()
+	if err == nil || !strings.Contains(err.Error(), "cycle") {
+		t.Fatalf("want cycle error, got %v", err)
+	}
+}
+
+func jobBlock(name, dep string) string {
+	return "  " + name + ":\n    command: convert\n    depends_on: [" + dep + "]\n" +
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n" +
+		"    output:\n      format: json\n      path: " + name + ".json\n"
+}
+
+func TestPlanTopologicalOrder(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		"  base:\n    command: convert\n    inputs:\n      - path: a.dbml\n        format: dbml\n    output:\n      format: json\n      path: base.json\n" +
+		"  mid:\n    command: convert\n    depends_on: [base]\n    inputs:\n      - path: a.dbml\n        format: dbml\n    output:\n      format: json\n      path: mid.json\n" +
+		"  top:\n    command: convert\n    depends_on: [mid]\n    inputs:\n      - path: a.dbml\n        format: dbml\n    output:\n      format: json\n      path: top.json\n"
+	set := loadOne(t, content)
+	if err := set.Validate(); err != nil {
+		t.Fatalf("validate: %v", err)
+	}
+	plan, err := set.Plan("top", true)
+	if err != nil {
+		t.Fatal(err)
+	}
+	var order []string
+	for _, j := range plan {
+		order = append(order, j.Name)
+	}
+	if strings.Join(order, ",") != "base,mid,top" {
+		t.Fatalf("plan order = %v, want [base mid top]", order)
+	}
+
+	solo, err := set.Plan("top", false)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(solo) != 1 || solo[0].Name != "top" {
+		t.Fatalf("no-deps plan = %v, want [top]", solo)
+	}
+}
+
+func TestSafeJoinStaysInsideRoot(t *testing.T) {
+	root := t.TempDir()
+	if _, err := SafeJoin(root, "sub/dir/file.sql"); err != nil {
+		t.Fatalf("expected ok, got %v", err)
+	}
+	if _, err := SafeJoin(root, "../escape"); err == nil {
+		t.Fatal("expected escape rejection")
+	}
+	if _, err := SafeJoin(root, "/abs"); err == nil {
+		t.Fatal("expected absolute rejection")
+	}
+}
+
+func TestShippedExampleIsValid(t *testing.T) {
+	path := filepath.Join("..", "..", "examples", "jobs", "relspec.yml")
+	set, err := Load([]string{path})
+	if err != nil {
+		t.Fatalf("load example: %v", err)
+	}
+	if err := set.Validate(); err != nil {
+		t.Fatalf("example manifest failed validation: %v", err)
+	}
+	if _, err := set.Plan("build-json", true); err != nil {
+		t.Fatalf("plan example: %v", err)
+	}
+}
+
+func TestFromJobWiring(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		"  producer:\n    command: convert\n" +
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n" +
+		"    output:\n      format: json\n      path: build/schema.json\n" +
+		"  consumer:\n    command: convert\n" +
+		"    inputs:\n      - from_job: producer\n" +
+		"    output:\n      format: yaml\n      path: build/schema.yaml\n"
+	set := loadOne(t, content)
+	if err := set.Validate(); err != nil {
+		t.Fatalf("validate: %v", err)
+	}
+	plan, err := set.Plan("consumer", true)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(plan) != 2 || plan[0].Name != "producer" || plan[1].Name != "consumer" {
+		t.Fatalf("plan = %v, want [producer consumer]", plan)
+	}
+}
+
+func TestFromJobRejectsNonProducer(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		"  lister:\n    command: scripts-list\n    script_dirs: [migrations]\n" +
+		"  consumer:\n    command: convert\n" +
+		"    inputs:\n      - from_job: lister\n" +
+		"    output:\n      format: yaml\n      path: out.yaml\n"
+	set := loadOne(t, content)
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "from_job") {
+		t.Fatalf("want from_job producer error, got %v", err)
+	}
+}
+
+func TestFromJobRejectsUnknownJob(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		"  consumer:\n    command: convert\n" +
+		"    inputs:\n      - from_job: ghost\n" +
+		"    output:\n      format: yaml\n      path: out.yaml\n"
+	set := loadOne(t, content)
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "unknown job") {
+		t.Fatalf("want unknown job error, got %v", err)
+	}
+}
+
+func TestFromJobCycleDetected(t *testing.T) {
+	content := "version: 1\njobs:\n" +
+		"  a:\n    command: convert\n" +
+		"    inputs:\n      - from_job: b\n" +
+		"    output:\n      format: json\n      path: a.json\n" +
+		"  b:\n    command: convert\n" +
+		"    inputs:\n      - from_job: a\n" +
+		"    output:\n      format: json\n      path: b.json\n"
+	set := loadOne(t, content)
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "cycle") {
+		t.Fatalf("want cycle error, got %v", err)
+	}
+}
+
+func TestSplitJobValidation(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  s:\n    command: split\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "missing output") {
+		t.Fatalf("want missing output, got %v", err)
+	}
+	set = loadOne(t, "version: 1\njobs:\n  s:\n    command: split\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    select:\n      tables: [users]\n"+
+		"    output:\n      format: json\n      path: out.json\n")
+	if err := set.Validate(); err != nil {
+		t.Fatalf("expected valid split job, got %v", err)
+	}
+}
+
+func TestInspectJobValidation(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  i:\n    command: inspect\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "report") {
+		t.Fatalf("want report required, got %v", err)
+	}
+	set = loadOne(t, "version: 1\njobs:\n  i:\n    command: inspect\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    report:\n      format: json\n      path: build/report.json\n")
+	if err := set.Validate(); err != nil {
+		t.Fatalf("expected valid inspect job, got %v", err)
+	}
+}
+
+func TestDiffJobValidation(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  d:\n    command: diff\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"    report:\n      format: summary\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "exactly 2 inputs") {
+		t.Fatalf("want exactly 2 inputs, got %v", err)
+	}
+	set = loadOne(t, "version: 1\njobs:\n  d:\n    command: diff\n"+
+		"    inputs:\n      - path: a.dbml\n        format: dbml\n"+
+		"      - path: b.dbml\n        format: dbml\n"+
+		"    report:\n      format: summary\n")
+	if err := set.Validate(); err != nil {
+		t.Fatalf("expected valid diff job, got %v", err)
+	}
+}
+
+func TestScriptsExecValidation(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  x:\n    command: scripts-exec\n"+
+		"    script_dirs: [migrations]\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "conn_env") {
+		t.Fatalf("want output.conn_env required, got %v", err)
+	}
+	set = loadOne(t, "version: 1\njobs:\n  x:\n    command: scripts-exec\n"+
+		"    script_dirs: [migrations]\n"+
+		"    output:\n      conn_env: TARGET_DB_URL\n")
+	if err := set.Validate(); err != nil {
+		t.Fatalf("expected valid scripts-exec job, got %v", err)
+	}
+}
+
+func TestSafeJoinRejectsSymlinkEscape(t *testing.T) {
+	root := t.TempDir()
+	outside := t.TempDir()
+	link := filepath.Join(root, "link")
+	if err := os.Symlink(outside, link); err != nil {
+		t.Skipf("symlink not supported: %v", err)
+	}
+	if _, err := SafeJoin(root, "link/x.sql"); err == nil {
+		t.Fatal("expected rejection of a path escaping via a symlink")
+	}
+}
+
+func TestScriptsListValidation(t *testing.T) {
+	set := loadOne(t, "version: 1\njobs:\n  s:\n    command: scripts-list\n")
+	if err := set.Validate(); err == nil || !strings.Contains(err.Error(), "script_dir") {
+		t.Fatalf("want script_dir required error, got %v", err)
+	}
+	set = loadOne(t, "version: 1\njobs:\n  s:\n    command: scripts-list\n    script_dirs: [migrations, extra]\n")
+	if err := set.Validate(); err != nil {
+		t.Fatalf("expected valid scripts-list job, got %v", err)
+	}
+}

pkg/merge/merge.go

@@ -118,7 +118,7 @@ func (r *MergeResult) mergeSchemaContents(target, source *models.Schema, opts *M
 	}
 }
 
-func (r *MergeResult) mergeTables(schema *models.Schema, source *models.Schema, opts *MergeOptions) {
+func (r *MergeResult) mergeTables(schema, source *models.Schema, opts *MergeOptions) {
 	// Create map of existing tables
 	existingTables := make(map[string]*models.Table)
 	for _, table := range schema.Tables {
@@ -150,7 +150,7 @@ func (r *MergeResult) mergeTables(schema *models.Schema, source *models.Schema,
 	}
 }
 
-func (r *MergeResult) mergeColumns(table *models.Table, srcTable *models.Table) {
+func (r *MergeResult) mergeColumns(table, srcTable *models.Table) {
 	// Create map of existing columns
 	existingColumns := make(map[string]*models.Column)
 	for colName := range table.Columns {
@@ -185,7 +185,7 @@ func (r *MergeResult) mergeColumns(table *models.Table, srcTable *models.Table)
 	}
 }
 
-func (r *MergeResult) mergeConstraints(table *models.Table, srcTable *models.Table) {
+func (r *MergeResult) mergeConstraints(table, srcTable *models.Table) {
 	// Initialize constraints map if nil
 	if table.Constraints == nil {
 		table.Constraints = make(map[string]*models.Constraint)
@@ -208,7 +208,7 @@ func (r *MergeResult) mergeConstraints(table *models.Table, srcTable *models.Tab
 	}
 }
 
-func (r *MergeResult) mergeIndexes(table *models.Table, srcTable *models.Table) {
+func (r *MergeResult) mergeIndexes(table, srcTable *models.Table) {
 	// Initialize indexes map if nil
 	if table.Indexes == nil {
 		table.Indexes = make(map[string]*models.Index)
@@ -231,7 +231,7 @@ func (r *MergeResult) mergeIndexes(table *models.Table, srcTable *models.Table)
 	}
 }
 
-func (r *MergeResult) mergeViews(schema *models.Schema, source *models.Schema) {
+func (r *MergeResult) mergeViews(schema, source *models.Schema) {
 	// Create map of existing views
 	existingViews := make(map[string]*models.View)
 	for _, view := range schema.Views {
@@ -250,7 +250,7 @@ func (r *MergeResult) mergeViews(schema *models.Schema, source *models.Schema) {
 	}
 }
 
-func (r *MergeResult) mergeSequences(schema *models.Schema, source *models.Schema) {
+func (r *MergeResult) mergeSequences(schema, source *models.Schema) {
 	// Create map of existing sequences
 	existingSequences := make(map[string]*models.Sequence)
 	for _, seq := range schema.Sequences {
@@ -269,7 +269,7 @@ func (r *MergeResult) mergeSequences(schema *models.Schema, source *models.Schem
 	}
 }
 
-func (r *MergeResult) mergeEnums(schema *models.Schema, source *models.Schema) {
+func (r *MergeResult) mergeEnums(schema, source *models.Schema) {
 	// Create map of existing enums
 	existingEnums := make(map[string]*models.Enum)
 	for _, enum := range schema.Enums {
@@ -288,7 +288,7 @@ func (r *MergeResult) mergeEnums(schema *models.Schema, source *models.Schema) {
 	}
 }
 
-func (r *MergeResult) mergeRelations(schema *models.Schema, source *models.Schema) {
+func (r *MergeResult) mergeRelations(schema, source *models.Schema) {
 	// Create map of existing relations
 	existingRelations := make(map[string]*models.Relationship)
 	for _, rel := range schema.Relations {
@@ -306,7 +306,7 @@ func (r *MergeResult) mergeRelations(schema *models.Schema, source *models.Schem
 	}
 }
 
-func (r *MergeResult) mergeDomains(target *models.Database, source *models.Database) {
+func (r *MergeResult) mergeDomains(target, source *models.Database) {
 	// Create map of existing domains
 	existingDomains := make(map[string]*models.Domain)
 	for _, domain := range target.Domains {

pkg/models/directives.go

@@ -0,0 +1,237 @@
+package models
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+)
+
+// Directive is a dialect-specific instruction embedded in a source schema
+// (currently DBML) that is preserved losslessly in the intermediate model and
+// consumed only by the writer for its namespace. Directives are stored in the
+// Metadata map of the object they apply to, under DirectivesMetadataKey.
+//
+// Example DBML: `@postgres: partition by RANGE (created_at)` parses to
+// Directive{Namespace: "postgres", Key: "partition", Args: "partition by RANGE (created_at)"}.
+type Directive struct {
+	// Namespace is the dialect the directive targets, e.g. "postgres" or "sqlite".
+	Namespace string `json:"namespace" yaml:"namespace"`
+	// Key is the lowercased first token of Args, used for duplicate detection
+	// and writer dispatch.
+	Key string `json:"key,omitempty" yaml:"key,omitempty"`
+	// Args is the verbatim argument text following the "@namespace:" prefix.
+	Args string `json:"args" yaml:"args"`
+	// Line is the 1-based source line the directive was read from, when known.
+	Line int `json:"line,omitempty" yaml:"line,omitempty"`
+}
+
+// DirectivesMetadataKey is the Metadata map key under which the ordered list of
+// dialect directives for an object is stored.
+const DirectivesMetadataKey = "directives"
+
+// DirectiveKey derives the Key for a directive from its argument text: the
+// lowercased first whitespace-delimited token.
+func DirectiveKey(args string) string {
+	fields := strings.Fields(args)
+	if len(fields) == 0 {
+		return ""
+	}
+	return strings.ToLower(fields[0])
+}
+
+// AddDirective appends d to the directive list stored in meta. The caller is
+// responsible for ensuring meta is non-nil (all Init* constructors allocate it).
+// If d.Key is empty it is derived from d.Args.
+func AddDirective(meta map[string]any, d Directive) {
+	if meta == nil {
+		return
+	}
+	if d.Key == "" {
+		d.Key = DirectiveKey(d.Args)
+	}
+	existing := GetDirectives(meta)
+	existing = append(existing, d)
+	meta[DirectivesMetadataKey] = existing
+}
+
+// GetDirectives returns the directives stored in meta, sorted deterministically
+// by (Namespace, Line, Args). It tolerates both a freshly built []Directive and
+// the []any of map[string]any produced by a JSON/YAML round-trip.
+func GetDirectives(meta map[string]any) []Directive {
+	if meta == nil {
+		return nil
+	}
+	raw, ok := meta[DirectivesMetadataKey]
+	if !ok || raw == nil {
+		return nil
+	}
+
+	var out []Directive
+	switch v := raw.(type) {
+	case []Directive:
+		out = append(out, v...)
+	case []any:
+		for _, item := range v {
+			if d, ok := directiveFromAny(item); ok {
+				out = append(out, d)
+			}
+		}
+	}
+
+	sort.SliceStable(out, func(i, j int) bool {
+		if out[i].Namespace != out[j].Namespace {
+			return out[i].Namespace < out[j].Namespace
+		}
+		if out[i].Line != out[j].Line {
+			return out[i].Line < out[j].Line
+		}
+		return out[i].Args < out[j].Args
+	})
+	return out
+}
+
+// directiveFromAny decodes a single directive from the loosely typed forms that
+// survive a JSON or YAML round-trip (map[string]any / map[any]any).
+func directiveFromAny(item any) (Directive, bool) {
+	switch m := item.(type) {
+	case Directive:
+		return m, true
+	case map[string]any:
+		return directiveFromStringMap(m), true
+	case map[any]any:
+		sm := make(map[string]any, len(m))
+		for k, val := range m {
+			if ks, ok := k.(string); ok {
+				sm[ks] = val
+			}
+		}
+		return directiveFromStringMap(sm), true
+	}
+	return Directive{}, false
+}
+
+func directiveFromStringMap(m map[string]any) Directive {
+	d := Directive{}
+	if s, ok := m["namespace"].(string); ok {
+		d.Namespace = s
+	}
+	if s, ok := m["key"].(string); ok {
+		d.Key = s
+	}
+	if s, ok := m["args"].(string); ok {
+		d.Args = s
+	}
+	switch n := m["line"].(type) {
+	case int:
+		d.Line = n
+	case int64:
+		d.Line = int(n)
+	case float64:
+		d.Line = int(n)
+	}
+	if d.Key == "" {
+		d.Key = DirectiveKey(d.Args)
+	}
+	return d
+}
+
+// DirectivesForNamespace returns the directives in meta that target ns, in the
+// deterministic order of GetDirectives.
+func DirectivesForNamespace(meta map[string]any, ns string) []Directive {
+	all := GetDirectives(meta)
+	if len(all) == 0 {
+		return nil
+	}
+	out := make([]Directive, 0, len(all))
+	for _, d := range all {
+		if d.Namespace == ns {
+			out = append(out, d)
+		}
+	}
+	return out
+}
+
+// HasDirective reports whether meta contains a directive with the given
+// namespace and key.
+func HasDirective(meta map[string]any, ns, key string) bool {
+	for _, d := range GetDirectives(meta) {
+		if d.Namespace == ns && d.Key == key {
+			return true
+		}
+	}
+	return false
+}
+
+// DirectiveSpec describes a documented directive in the catalog.
+type DirectiveSpec struct {
+	// Singleton means only one directive with this namespace/key may appear at
+	// a single location; a second one is a parse error.
+	Singleton bool
+	// Locations lists the location kinds the directive is valid at
+	// ("database", "table", "column", "index").
+	Locations []string
+}
+
+// Location kinds a directive may attach to.
+const (
+	DirectiveLocationDatabase = "database"
+	DirectiveLocationTable    = "table"
+	DirectiveLocationColumn   = "column"
+	DirectiveLocationIndex    = "index"
+)
+
+// DirectiveCatalog is the set of documented directives per namespace. It is used
+// for strict-mode validation in readers and writers; unknown namespaces/keys are
+// still preserved losslessly when strict mode is off.
+var DirectiveCatalog = map[string]map[string]DirectiveSpec{
+	"postgres": {
+		"partition":   {Singleton: true, Locations: []string{DirectiveLocationTable}},
+		"tablespace":  {Singleton: true, Locations: []string{DirectiveLocationTable, DirectiveLocationIndex}},
+		"inherits":    {Singleton: true, Locations: []string{DirectiveLocationTable}},
+		"with":        {Singleton: false, Locations: []string{DirectiveLocationTable, DirectiveLocationIndex}},
+		"storage":     {Singleton: true, Locations: []string{DirectiveLocationColumn}},
+		"compression": {Singleton: true, Locations: []string{DirectiveLocationColumn}},
+		"identity":    {Singleton: true, Locations: []string{DirectiveLocationColumn}},
+	},
+	"sqlite": {
+		"without": {Singleton: true, Locations: []string{DirectiveLocationTable}},
+		"strict":  {Singleton: true, Locations: []string{DirectiveLocationTable}},
+		"collate": {Singleton: true, Locations: []string{DirectiveLocationColumn}},
+	},
+}
+
+// LookupDirectiveSpec returns the catalog spec for a namespace/key and whether
+// it is documented.
+func LookupDirectiveSpec(ns, key string) (DirectiveSpec, bool) {
+	keys, ok := DirectiveCatalog[ns]
+	if !ok {
+		return DirectiveSpec{}, false
+	}
+	spec, ok := keys[key]
+	return spec, ok
+}
+
+// DirectiveLocationAllowed reports whether a documented directive may appear at
+// the given location. Unknown directives (not in the catalog) are allowed
+// everywhere so they can be preserved.
+func DirectiveLocationAllowed(ns, key, location string) bool {
+	spec, ok := LookupDirectiveSpec(ns, key)
+	if !ok {
+		return true
+	}
+	for _, l := range spec.Locations {
+		if l == location {
+			return true
+		}
+	}
+	return false
+}
+
+// FormatDirectiveLine renders a directive back to its DBML source form, e.g.
+// "@postgres: partition by RANGE (created_at)" or "@postgres(id): identity always".
+func FormatDirectiveLine(d Directive, target string) string {
+	if target != "" {
+		return fmt.Sprintf("@%s(%s): %s", d.Namespace, target, d.Args)
+	}
+	return fmt.Sprintf("@%s: %s", d.Namespace, d.Args)
+}

pkg/models/directives_test.go

@@ -0,0 +1,133 @@
+package models
+
+import (
+	"encoding/json"
+	"testing"
+)
+
+func TestDirectiveKey(t *testing.T) {
+	cases := map[string]string{
+		"partition by RANGE (created_at)": "partition",
+		"WITHOUT ROWID":                   "without",
+		"  strict  ":                      "strict",
+		"":                                "",
+	}
+	for args, want := range cases {
+		if got := DirectiveKey(args); got != want {
+			t.Errorf("DirectiveKey(%q) = %q, want %q", args, got, want)
+		}
+	}
+}
+
+func TestAddDirectiveDerivesKey(t *testing.T) {
+	meta := map[string]any{}
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "partition by RANGE (x)", Line: 2})
+	AddDirective(meta, Directive{Namespace: "postgres", Key: "tablespace", Args: "tablespace fast", Line: 3})
+
+	got := GetDirectives(meta)
+	if len(got) != 2 {
+		t.Fatalf("got %d directives, want 2", len(got))
+	}
+	if got[0].Key != "partition" {
+		t.Errorf("derived key = %q, want %q", got[0].Key, "partition")
+	}
+	if got[1].Key != "tablespace" {
+		t.Errorf("explicit key = %q, want %q", got[1].Key, "tablespace")
+	}
+}
+
+func TestAddDirectiveNilMeta(t *testing.T) {
+	// Must not panic.
+	AddDirective(nil, Directive{Namespace: "postgres", Args: "strict"})
+}
+
+func TestGetDirectivesOrdering(t *testing.T) {
+	meta := map[string]any{}
+	AddDirective(meta, Directive{Namespace: "sqlite", Args: "strict", Line: 9})
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "with (b)", Line: 5})
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "with (a)", Line: 5})
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "partition by x", Line: 2})
+
+	got := GetDirectives(meta)
+	wantArgs := []string{"partition by x", "with (a)", "with (b)", "strict"}
+	if len(got) != len(wantArgs) {
+		t.Fatalf("got %d directives, want %d", len(got), len(wantArgs))
+	}
+	for i, w := range wantArgs {
+		if got[i].Args != w {
+			t.Errorf("directive[%d].Args = %q, want %q", i, got[i].Args, w)
+		}
+	}
+}
+
+func TestGetDirectivesTolerantDecodeAfterJSON(t *testing.T) {
+	meta := map[string]any{}
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "partition by RANGE (created_at)", Line: 4})
+	AddDirective(meta, Directive{Namespace: "sqlite", Args: "without rowid", Line: 6})
+
+	blob, err := json.Marshal(meta)
+	if err != nil {
+		t.Fatalf("marshal: %v", err)
+	}
+	var round map[string]any
+	if err := json.Unmarshal(blob, &round); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+
+	got := GetDirectives(round)
+	if len(got) != 2 {
+		t.Fatalf("got %d directives after JSON round-trip, want 2", len(got))
+	}
+	if got[0].Namespace != "postgres" || got[0].Key != "partition" || got[0].Line != 4 {
+		t.Errorf("post-JSON directive[0] = %+v", got[0])
+	}
+	if got[0].Args != "partition by RANGE (created_at)" {
+		t.Errorf("post-JSON args not verbatim: %q", got[0].Args)
+	}
+	if got[1].Namespace != "sqlite" || got[1].Key != "without" {
+		t.Errorf("post-JSON directive[1] = %+v", got[1])
+	}
+}
+
+func TestDirectivesForNamespaceAndHasDirective(t *testing.T) {
+	meta := map[string]any{}
+	AddDirective(meta, Directive{Namespace: "postgres", Args: "partition by x", Line: 1})
+	AddDirective(meta, Directive{Namespace: "sqlite", Args: "strict", Line: 2})
+
+	pg := DirectivesForNamespace(meta, "postgres")
+	if len(pg) != 1 || pg[0].Key != "partition" {
+		t.Errorf("DirectivesForNamespace(postgres) = %+v", pg)
+	}
+	if !HasDirective(meta, "sqlite", "strict") {
+		t.Error("HasDirective(sqlite, strict) = false, want true")
+	}
+	if HasDirective(meta, "postgres", "tablespace") {
+		t.Error("HasDirective(postgres, tablespace) = true, want false")
+	}
+}
+
+func TestDirectiveLocationAllowed(t *testing.T) {
+	if !DirectiveLocationAllowed("postgres", "partition", DirectiveLocationTable) {
+		t.Error("partition should be allowed at table level")
+	}
+	if DirectiveLocationAllowed("postgres", "partition", DirectiveLocationColumn) {
+		t.Error("partition should not be allowed at column level")
+	}
+	// Unknown directives are allowed everywhere so they can be preserved.
+	if !DirectiveLocationAllowed("postgres", "bogus", DirectiveLocationDatabase) {
+		t.Error("unknown key should be allowed everywhere")
+	}
+	if !DirectiveLocationAllowed("madeup", "x", DirectiveLocationTable) {
+		t.Error("unknown namespace should be allowed everywhere")
+	}
+}
+
+func TestFormatDirectiveLine(t *testing.T) {
+	d := Directive{Namespace: "postgres", Key: "identity", Args: "identity always"}
+	if got := FormatDirectiveLine(d, ""); got != "@postgres: identity always" {
+		t.Errorf("FormatDirectiveLine no target = %q", got)
+	}
+	if got := FormatDirectiveLine(d, "id"); got != "@postgres(id): identity always" {
+		t.Errorf("FormatDirectiveLine with target = %q", got)
+	}
+}

pkg/models/models.go

@@ -32,6 +32,7 @@ type Database struct {
 	DatabaseType    DatabaseType   `json:"database_type,omitempty" yaml:"database_type,omitempty" xml:"database_type,omitempty"`
 	DatabaseVersion string         `json:"database_version,omitempty" yaml:"database_version,omitempty" xml:"database_version,omitempty"`
 	SourceFormat    string         `json:"source_format,omitempty" yaml:"source_format,omitempty" xml:"source_format,omitempty"` // Source Format of the database.
+	Metadata        map[string]any `json:"metadata,omitempty" yaml:"metadata,omitempty" xml:"-"`
 	UpdatedAt       string         `json:"updatedat,omitempty" yaml:"updatedat,omitempty" xml:"updatedat,omitempty"`
 	GUID            string         `json:"guid" yaml:"guid" xml:"guid"`
 }
@@ -240,6 +241,7 @@ type Column struct {
 	IsPrimaryKey  bool           `json:"is_primary_key" yaml:"is_primary_key" xml:"is_primary_key"`
 	Comment       string         `json:"comment,omitempty" yaml:"comment,omitempty" xml:"comment,omitempty"`
 	Collation     string         `json:"collation,omitempty" yaml:"collation,omitempty" xml:"collation,omitempty"`
+	Metadata      map[string]any `json:"metadata,omitempty" yaml:"metadata,omitempty" xml:"-"`
 	Sequence      uint           `json:"sequence,omitempty" yaml:"sequence,omitempty" xml:"sequence,omitempty"`
 	GUID          string         `json:"guid" yaml:"guid" xml:"guid"`
 }
@@ -263,6 +265,7 @@ type Index struct {
 	Concurrent  bool           `json:"concurrent,omitempty" yaml:"concurrent,omitempty" xml:"concurrent,omitempty"`
 	Include     []string       `json:"include,omitempty" yaml:"include,omitempty" xml:"include,omitempty"` // INCLUDE columns
 	Comment     string         `json:"comment,omitempty" yaml:"comment,omitempty" xml:"comment,omitempty"`
+	Metadata    map[string]any `json:"metadata,omitempty" yaml:"metadata,omitempty" xml:"-"`
 	Sequence    uint           `json:"sequence,omitempty" yaml:"sequence,omitempty" xml:"sequence,omitempty"`
 	GUID        string         `json:"guid" yaml:"guid" xml:"guid"`
 }
@@ -396,6 +399,7 @@ func InitDatabase(name string) *Database {
 		Name:     name,
 		Schemas:  make([]*Schema, 0),
 		Domains:  make([]*Domain, 0),
+		Metadata: make(map[string]any),
 		GUID:     uuid.New().String(),
 	}
 }
@@ -434,6 +438,7 @@ func InitColumn(name, table, schema string) *Column {
 		Name:     name,
 		Table:    table,
 		Schema:   schema,
+		Metadata: make(map[string]any),
 		GUID:     uuid.New().String(),
 	}
 }
@@ -446,6 +451,7 @@ func InitIndex(name, table, schema string) *Index {
 		Schema:   schema,
 		Columns:  make([]string, 0),
 		Include:  make([]string, 0),
+		Metadata: make(map[string]any),
 		GUID:     uuid.New().String(),
 	}
 }

pkg/pgsql/extensions.go

@@ -0,0 +1,460 @@
+package pgsql
+
+import (
+	"sort"
+	"strings"
+)
+
+// Extension describes a PostgreSQL extension RelSpec recognizes, along with the schema
+// artefacts that imply it: the types it provides (declared on TypeSpec.Extension), the
+// index access methods and operator classes it installs, and the functions whose use in a
+// default, check constraint, index predicate, or view body requires it.
+type Extension struct {
+	Name        string
+	Category    string
+	Description string
+
+	// Requires lists extensions that must be created before this one.
+	Requires []string
+
+	// IndexMethods are access methods usable as Index.Type.
+	IndexMethods []string
+
+	// OperatorClasses are operator classes the extension installs.
+	OperatorClasses []string
+
+	// Functions are function names whose use implies the extension.
+	Functions []string
+
+	// FunctionPrefixes match whole families of functions (e.g. "st_" for PostGIS).
+	FunctionPrefixes []string
+}
+
+// postgresExtensions is the set of extensions RelSpec knows how to detect and emit.
+var postgresExtensions = map[string]Extension{
+	"amcheck": {
+		Name: "amcheck", Category: "integrity",
+		Description: "Verifies B-tree and related structure consistency to help detect corruption.",
+		Functions:   []string{"bt_index_check", "bt_index_parent_check", "verify_heapam"},
+	},
+	"btree_gin": {
+		Name: "btree_gin", Category: "indexing",
+		Description: "Adds GIN operator classes for common scalar data types.",
+	},
+	"btree_gist": {
+		Name: "btree_gist", Category: "indexing",
+		Description: "Adds GiST operator classes for common scalar data types and exclusion constraints.",
+	},
+	"citext": {
+		Name: "citext", Category: "text",
+		Description: "Provides case-insensitive text columns and operators.",
+		Functions:   []string{"citext"},
+	},
+	"fuzzystrmatch": {
+		Name: "fuzzystrmatch", Category: "text",
+		Description: "Adds phonetic and fuzzy matching helpers like Soundex and Levenshtein.",
+		Functions: []string{
+			"soundex", "difference", "levenshtein", "levenshtein_less_equal",
+			"metaphone", "dmetaphone", "dmetaphone_alt",
+		},
+	},
+	"hstore": {
+		Name: "hstore", Category: "document",
+		Description:     "Adds a lightweight key/value data type for semi-structured attributes.",
+		OperatorClasses: []string{"gin_hstore_ops", "gist_hstore_ops", "hash_hstore_ops", "btree_hstore_ops"},
+		Functions: []string{
+			"hstore", "akeys", "avals", "skeys", "svals",
+			"hstore_to_json", "hstore_to_jsonb", "hstore_to_array", "hstore_to_matrix",
+		},
+	},
+	"http": {
+		Name: "http", Category: "integration",
+		Description: "Lets SQL functions make outbound HTTP requests.",
+		Functions: []string{
+			"http", "http_get", "http_post", "http_put", "http_patch", "http_delete",
+			"http_head", "urlencode",
+		},
+	},
+	"pg_background": {
+		Name: "pg_background", Category: "jobs",
+		Description: "Runs SQL asynchronously in PostgreSQL background workers.",
+		Functions:   []string{"pg_background_launch", "pg_background_result", "pg_background_detach"},
+	},
+	"pg_cron": {
+		Name: "pg_cron", Category: "scheduling",
+		Description:      "Schedules recurring SQL jobs inside PostgreSQL.",
+		FunctionPrefixes: []string{"cron."},
+	},
+	"pg_jsonschema": {
+		Name: "pg_jsonschema", Category: "validation",
+		Description: "Validates json and jsonb values against JSON Schema.",
+		Functions:   []string{"json_matches_schema", "jsonb_matches_schema", "jsonschema_is_valid"},
+	},
+	"pg_partman": {
+		Name: "pg_partman", Category: "partitioning",
+		Description:      "Automates time-based and serial-based partition management.",
+		FunctionPrefixes: []string{"partman."},
+	},
+	"pg_qualstats": {
+		Name: "pg_qualstats", Category: "observability",
+		Description: "Tracks predicate usage in WHERE and JOIN clauses for tuning and index advice.",
+	},
+	"pg_repack": {
+		Name: "pg_repack", Category: "maintenance",
+		Description: "Rebuilds bloated tables and indexes online with minimal locking.",
+	},
+	"pg_search": {
+		Name: "pg_search", Category: "search",
+		Description: "Provides ParadeDB full-text and relevance search features.",
+		// bm25 is also the access method name used by pg_textsearch; pg_search is the
+		// canonical provider, so a bm25 index resolves to it.
+		IndexMethods:     []string{"bm25"},
+		FunctionPrefixes: []string{"paradedb."},
+	},
+	"pg_stat_statements": {
+		Name: "pg_stat_statements", Category: "observability",
+		Description: "Tracks normalized query execution statistics.",
+	},
+	"pg_textsearch": {
+		Name: "pg_textsearch", Category: "search",
+		Description: "Adds BM25-style text search support.",
+	},
+	"pg_trgm": {
+		Name: "pg_trgm", Category: "text",
+		Description:     "Adds trigram similarity search and fast fuzzy matching indexes.",
+		OperatorClasses: []string{"gin_trgm_ops", "gist_trgm_ops"},
+		Functions: []string{
+			"similarity", "word_similarity", "strict_word_similarity",
+			"show_trgm", "show_limit", "set_limit",
+		},
+	},
+	"pgcrypto": {
+		Name: "pgcrypto", Category: "security",
+		Description: "Adds hashing, encryption, random bytes, and UUID helpers.",
+		// gen_random_uuid is deliberately absent: it is built in since PostgreSQL 13.
+		Functions: []string{
+			"crypt", "gen_salt", "gen_random_bytes", "digest", "hmac",
+			"pgp_sym_encrypt", "pgp_sym_decrypt", "pgp_pub_encrypt", "pgp_pub_decrypt",
+			"armor", "dearmor",
+		},
+	},
+	"pgrouting": {
+		Name: "pgrouting", Category: "geospatial",
+		Description:      "Adds routing and graph algorithms on top of PostGIS data.",
+		Requires:         []string{"postgis"},
+		FunctionPrefixes: []string{"pgr_"},
+	},
+	"pgstattuple": {
+		Name: "pgstattuple", Category: "maintenance",
+		Description: "Reports table and index tuple density and bloat information.",
+		Functions:   []string{"pgstattuple", "pgstatindex", "pgstatginindex", "pg_relpages"},
+	},
+	"plpython3u": {
+		Name: "plpython3u", Category: "procedural",
+		Description: "Lets you write PostgreSQL functions in Python 3.",
+	},
+	"postgis": {
+		Name: "postgis", Category: "geospatial",
+		Description:  "Adds spatial data types, functions, and indexes.",
+		IndexMethods: nil, // uses the built-in gist/spgist/brin access methods
+		OperatorClasses: []string{
+			"gist_geometry_ops_2d", "gist_geometry_ops_nd", "gist_geography_ops",
+			"spgist_geometry_ops_2d", "spgist_geometry_ops_3d", "spgist_geometry_ops_nd",
+			"brin_geometry_inclusion_ops_2d", "brin_geometry_inclusion_ops_3d",
+			"brin_geometry_inclusion_ops_4d", "brin_geography_inclusion_ops_2d",
+			"btree_geometry_ops", "btree_geography_ops",
+		},
+		FunctionPrefixes: []string{"st_"},
+		Functions: []string{
+			"geometrytype", "addgeometrycolumn", "dropgeometrycolumn", "updategeometrysrid",
+			"find_srid", "postgis_version", "postgis_full_version",
+		},
+	},
+	"postgis_raster": {
+		Name: "postgis_raster", Category: "geospatial",
+		Description: "Adds the raster type and raster analysis functions.",
+		Requires:    []string{"postgis"},
+	},
+	"postgis_topology": {
+		Name: "postgis_topology", Category: "geospatial",
+		Description:      "Adds topology-aware spatial models and validation tools.",
+		Requires:         []string{"postgis"},
+		FunctionPrefixes: []string{"topology."},
+	},
+	"postgres_fdw": {
+		Name: "postgres_fdw", Category: "federation",
+		Description: "Connects PostgreSQL tables to other PostgreSQL servers.",
+	},
+	"timescaledb": {
+		Name: "timescaledb", Category: "time-series",
+		Description: "Adds hypertables, compression, retention, and time-series optimizations.",
+		Functions: []string{
+			"create_hypertable", "add_dimension", "time_bucket", "time_bucket_gapfill",
+			"add_retention_policy", "add_compression_policy", "locf", "interpolate",
+		},
+	},
+	"unaccent": {
+		Name: "unaccent", Category: "text",
+		Description: "Removes accents and diacritics for normalized text search.",
+		Functions:   []string{"unaccent"},
+	},
+	"uuid-ossp": {
+		Name: "uuid-ossp", Category: "utility",
+		Description: "Generates UUIDs using several algorithms.",
+		Functions: []string{
+			"uuid_generate_v1", "uuid_generate_v1mc", "uuid_generate_v3",
+			"uuid_generate_v4", "uuid_generate_v5",
+			"uuid_nil", "uuid_ns_dns", "uuid_ns_url", "uuid_ns_oid", "uuid_ns_x500",
+		},
+	},
+	"vector": {
+		Name: "vector", Category: "ai/search",
+		Description:  "Adds vector data types and similarity search for embeddings.",
+		IndexMethods: []string{"hnsw", "ivfflat"},
+		OperatorClasses: []string{
+			"vector_l2_ops", "vector_ip_ops", "vector_cosine_ops", "vector_l1_ops",
+			"halfvec_l2_ops", "halfvec_ip_ops", "halfvec_cosine_ops", "halfvec_l1_ops",
+			"sparsevec_l2_ops", "sparsevec_ip_ops", "sparsevec_cosine_ops", "sparsevec_l1_ops",
+			"bit_hamming_ops", "bit_jaccard_ops",
+		},
+		Functions: []string{"l2_distance", "inner_product", "cosine_distance", "l1_distance", "vector_dims", "vector_norm"},
+	},
+	"vchord": {
+		Name: "vchord", Category: "ai/search",
+		Description:  "Adds VectorChord scalable disk-friendly vector indexes compatible with pgvector data types.",
+		Requires:     []string{"vector"},
+		IndexMethods: []string{"vchordrq", "vchordg"},
+	},
+	"ltree": {
+		Name: "ltree", Category: "document",
+		Description:     "Adds a hierarchical label tree type.",
+		OperatorClasses: []string{"gist_ltree_ops", "gin_ltree_ops", "gist__ltree_ops"},
+		Functions:       []string{"subltree", "subpath", "nlevel", "lca", "ltree2text", "text2ltree"},
+	},
+}
+
+// extensionIndexMethods maps an index access method to the extension providing it.
+var extensionIndexMethods = buildExtensionIndex(func(ext Extension) []string { return ext.IndexMethods })
+
+// extensionOperatorClasses maps an operator class to the extension providing it.
+var extensionOperatorClasses = buildExtensionIndex(func(ext Extension) []string { return ext.OperatorClasses })
+
+// extensionFunctions maps a function name to the extension providing it.
+var extensionFunctions = buildExtensionIndex(func(ext Extension) []string { return ext.Functions })
+
+// extensionFunctionPrefixes maps a function name prefix to the extension providing it.
+var extensionFunctionPrefixes = buildExtensionIndex(func(ext Extension) []string { return ext.FunctionPrefixes })
+
+func buildExtensionIndex(keys func(Extension) []string) map[string]string {
+	index := make(map[string]string)
+	for name := range postgresExtensions {
+		ext := postgresExtensions[name]
+		for _, key := range keys(ext) {
+			// Deterministic on collision: the alphabetically first extension wins.
+			if existing, ok := index[key]; ok && existing < ext.Name {
+				continue
+			}
+			index[key] = ext.Name
+		}
+	}
+	return index
+}
+
+// LookupExtension returns the registered extension by name.
+func LookupExtension(name string) (Extension, bool) {
+	ext, ok := postgresExtensions[strings.ToLower(strings.TrimSpace(name))]
+	return ext, ok
+}
+
+// IsKnownExtension reports whether the named extension is registered.
+func IsKnownExtension(name string) bool {
+	_, ok := LookupExtension(name)
+	return ok
+}
+
+// GetExtensions returns every registered extension name, sorted.
+func GetExtensions() []string {
+	names := make([]string, 0, len(postgresExtensions))
+	for name := range postgresExtensions {
+		names = append(names, name)
+	}
+	sort.Strings(names)
+	return names
+}
+
+// IndexMethodExtension returns the extension providing an index access method
+// ("hnsw" -> "vector", "vchordrq" -> "vchord"). Built-in methods return "".
+func IndexMethodExtension(method string) string {
+	return extensionIndexMethods[strings.ToLower(strings.TrimSpace(method))]
+}
+
+// OperatorClassExtension returns the extension providing an operator class
+// ("gin_trgm_ops" -> "pg_trgm"). Built-in operator classes return "".
+func OperatorClassExtension(opClass string) string {
+	return extensionOperatorClasses[strings.ToLower(strings.TrimSpace(opClass))]
+}
+
+// ExtensionsForExpression returns the extensions whose functions appear in a SQL
+// expression such as a column default, check constraint, index predicate, or view body.
+// The result is sorted and deduplicated.
+func ExtensionsForExpression(expression string) []string {
+	if strings.TrimSpace(expression) == "" {
+		return nil
+	}
+
+	lower := strings.ToLower(expression)
+	found := make(map[string]bool)
+
+	for _, call := range sqlFunctionCalls(lower) {
+		if ext, ok := extensionFunctions[call]; ok {
+			found[ext] = true
+			continue
+		}
+		for prefix, ext := range extensionFunctionPrefixes {
+			if strings.HasPrefix(call, prefix) {
+				found[ext] = true
+				break
+			}
+		}
+	}
+
+	if len(found) == 0 {
+		return nil
+	}
+
+	names := make([]string, 0, len(found))
+	for name := range found {
+		names = append(names, name)
+	}
+	sort.Strings(names)
+	return names
+}
+
+// sqlFunctionCalls returns the lowercase names of every function call in an expression.
+// A call is an identifier (optionally schema-qualified) immediately followed by "(".
+func sqlFunctionCalls(lowerExpression string) []string {
+	calls := make([]string, 0, 4)
+	end := 0
+
+	for i := 0; i < len(lowerExpression); i++ {
+		if lowerExpression[i] != '(' {
+			continue
+		}
+
+		end = i
+		// Allow whitespace between the identifier and its opening parenthesis.
+		for end > 0 && isSQLSpace(lowerExpression[end-1]) {
+			end--
+		}
+
+		start := end
+		for start > 0 && isSQLIdentifierByte(lowerExpression[start-1]) {
+			start--
+		}
+		if start == end {
+			continue
+		}
+		// A leading digit means this is not an identifier (e.g. "2(").
+		if lowerExpression[start] >= '0' && lowerExpression[start] <= '9' {
+			continue
+		}
+		calls = append(calls, lowerExpression[start:end])
+	}
+
+	return calls
+}
+
+func isSQLIdentifierByte(b byte) bool {
+	switch {
+	case b >= 'a' && b <= 'z', b >= 'A' && b <= 'Z', b >= '0' && b <= '9':
+		return true
+	case b == '_', b == '.':
+		return true
+	default:
+		return false
+	}
+}
+
+func isSQLSpace(b byte) bool {
+	return b == ' ' || b == '\t' || b == '\n' || b == '\r'
+}
+
+// SortExtensions orders extension names so that dependencies come first (postgis before
+// postgis_topology, vector before vchord), with alphabetical order breaking ties.
+// Duplicates are removed; unknown names are kept and sorted alphabetically.
+func SortExtensions(names []string) []string {
+	unique := make(map[string]bool, len(names))
+	for _, name := range names {
+		name = strings.ToLower(strings.TrimSpace(name))
+		if name != "" {
+			unique[name] = true
+		}
+	}
+	if len(unique) == 0 {
+		return nil
+	}
+
+	pending := make([]string, 0, len(unique))
+	for name := range unique {
+		pending = append(pending, name)
+	}
+	sort.Strings(pending)
+
+	sorted := make([]string, 0, len(pending))
+	emitted := make(map[string]bool, len(pending))
+
+	var emit func(name string, seen map[string]bool)
+	emit = func(name string, seen map[string]bool) {
+		if emitted[name] || seen[name] {
+			return
+		}
+		seen[name] = true
+
+		if ext, ok := LookupExtension(name); ok {
+			for _, dependency := range ext.Requires {
+				// Only order dependencies that are actually being created.
+				if unique[dependency] {
+					emit(dependency, seen)
+				}
+			}
+		}
+
+		emitted[name] = true
+		sorted = append(sorted, name)
+	}
+
+	for _, name := range pending {
+		emit(name, make(map[string]bool))
+	}
+	return sorted
+}
+
+// ExtensionDependencies returns the extensions a given extension requires, sorted.
+func ExtensionDependencies(name string) []string {
+	ext, ok := LookupExtension(name)
+	if !ok || len(ext.Requires) == 0 {
+		return nil
+	}
+	requires := append([]string(nil), ext.Requires...)
+	sort.Strings(requires)
+	return requires
+}
+
+// QuoteExtensionName quotes an extension name when it is not a bare SQL identifier,
+// e.g. uuid-ossp -> "uuid-ossp".
+func QuoteExtensionName(name string) string {
+	name = strings.TrimSpace(name)
+	if name == "" {
+		return ""
+	}
+	for i := 0; i < len(name); i++ {
+		b := name[i]
+		switch {
+		case b >= 'a' && b <= 'z', b == '_':
+		case b >= '0' && b <= '9' && i > 0:
+		default:
+			return `"` + strings.ReplaceAll(name, `"`, `""`) + `"`
+		}
+	}
+	return name
+}

pkg/pgsql/extensions_test.go

@@ -0,0 +1,165 @@
+package pgsql
+
+import (
+	"reflect"
+	"strings"
+	"testing"
+)
+
+func TestExtensionRegistryConsistency(t *testing.T) {
+	for name, ext := range postgresExtensions {
+		if name != ext.Name {
+			t.Errorf("extension registered as %q has Name %q", name, ext.Name)
+		}
+		if name != strings.ToLower(name) {
+			t.Errorf("extension %q must be registered lowercase", name)
+		}
+		if ext.Description == "" || ext.Category == "" {
+			t.Errorf("extension %q is missing a category or description", name)
+		}
+		for _, dependency := range ext.Requires {
+			if !IsKnownExtension(dependency) {
+				t.Errorf("extension %q requires unregistered extension %q", name, dependency)
+			}
+		}
+	}
+}
+
+// Every extension named by a type in the type registry must itself be registered,
+// otherwise a column type would ask for a CREATE EXTENSION nothing knows how to order.
+func TestTypeExtensionsAreRegistered(t *testing.T) {
+	for typeName, spec := range postgresBaseTypes {
+		if spec.Extension == "" {
+			continue
+		}
+		if !IsKnownExtension(spec.Extension) {
+			t.Errorf("type %q declares unregistered extension %q", typeName, spec.Extension)
+		}
+	}
+}
+
+func TestIndexMethodExtension(t *testing.T) {
+	tests := map[string]string{
+		"hnsw":     "vector",
+		"ivfflat":  "vector",
+		"HNSW":     "vector",
+		"vchordrq": "vchord",
+		"vchordg":  "vchord",
+		"bm25":     "pg_search",
+		"btree":    "",
+		"gin":      "",
+		"":         "",
+	}
+
+	for method, want := range tests {
+		if got := IndexMethodExtension(method); got != want {
+			t.Errorf("IndexMethodExtension(%q) = %q, want %q", method, got, want)
+		}
+	}
+}
+
+func TestOperatorClassExtension(t *testing.T) {
+	tests := map[string]string{
+		"gin_trgm_ops":         "pg_trgm",
+		"gist_trgm_ops":        "pg_trgm",
+		"vector_cosine_ops":    "vector",
+		"halfvec_l2_ops":       "vector",
+		"gist_ltree_ops":       "ltree",
+		"gist_geometry_ops_2d": "postgis",
+		"jsonb_path_ops":       "",
+		"array_ops":            "",
+		"":                     "",
+	}
+
+	for opClass, want := range tests {
+		if got := OperatorClassExtension(opClass); got != want {
+			t.Errorf("OperatorClassExtension(%q) = %q, want %q", opClass, got, want)
+		}
+	}
+}
+
+func TestExtensionsForExpression(t *testing.T) {
+	tests := []struct {
+		name       string
+		expression string
+		want       []string
+	}{
+		{"empty", "", nil},
+		{"no functions", "status = 'active'", nil},
+		{"builtin only", "now()", nil},
+		{"uuid-ossp default", "uuid_generate_v4()", []string{"uuid-ossp"}},
+		{"gen_random_uuid is builtin", "gen_random_uuid()", nil},
+		{"pgcrypto", "crypt(password, gen_salt('bf'))", []string{"pgcrypto"}},
+		{"postgis prefix", "ST_Area(geom) > 0", []string{"postgis"}},
+		{"paradedb prefix", "paradedb.snippet(body)", []string{"pg_search"}},
+		{"jsonschema", "json_matches_schema('{}', payload)", []string{"pg_jsonschema"}},
+		{"whitespace before paren", "unaccent ('crème')", []string{"unaccent"}},
+		{"multiple sorted", "ST_X(geom) = levenshtein(a, b)::float", []string{"fuzzystrmatch", "postgis"}},
+		{"column named like function", "similarity_score > 0.5", nil},
+		{"numeric prefix ignored", "2(3)", nil},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := ExtensionsForExpression(tt.expression); !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("ExtensionsForExpression(%q) = %v, want %v", tt.expression, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestSortExtensions(t *testing.T) {
+	tests := []struct {
+		name  string
+		input []string
+		want  []string
+	}{
+		{"empty", nil, nil},
+		{"alphabetical", []string{"pg_trgm", "citext"}, []string{"citext", "pg_trgm"}},
+		{"deduplicated", []string{"vector", "vector", " VECTOR "}, []string{"vector"}},
+		{"dependency first", []string{"vchord", "vector"}, []string{"vector", "vchord"}},
+		{
+			"postgis dependants",
+			[]string{"postgis_topology", "pgrouting", "postgis"},
+			[]string{"postgis", "pgrouting", "postgis_topology"},
+		},
+		{"dependency not requested", []string{"vchord"}, []string{"vchord"}},
+		{"unknown names kept", []string{"zzz_custom", "citext"}, []string{"citext", "zzz_custom"}},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := SortExtensions(tt.input); !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("SortExtensions(%v) = %v, want %v", tt.input, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestQuoteExtensionName(t *testing.T) {
+	tests := map[string]string{
+		"vector":    "vector",
+		"pg_trgm":   "pg_trgm",
+		"uuid-ossp": `"uuid-ossp"`,
+		"PostGIS":   `"PostGIS"`,
+		"":          "",
+	}
+
+	for name, want := range tests {
+		if got := QuoteExtensionName(name); got != want {
+			t.Errorf("QuoteExtensionName(%q) = %q, want %q", name, got, want)
+		}
+	}
+}
+
+func TestExtensionDependencies(t *testing.T) {
+	if got := ExtensionDependencies("vchord"); !reflect.DeepEqual(got, []string{"vector"}) {
+		t.Errorf("ExtensionDependencies(vchord) = %v, want [vector]", got)
+	}
+	if got := ExtensionDependencies("citext"); got != nil {
+		t.Errorf("ExtensionDependencies(citext) = %v, want nil", got)
+	}
+	if got := ExtensionDependencies("not_an_extension"); got != nil {
+		t.Errorf("ExtensionDependencies(not_an_extension) = %v, want nil", got)
+	}
+}

pkg/pgsql/index_options.go

@@ -0,0 +1,248 @@
+package pgsql
+
+import (
+	"strconv"
+	"strings"
+)
+
+// Index access-method storage parameters, the WITH (...) clause of CREATE INDEX. RelSpec
+// carries them through the model in Index.Comment, so the parsing here is deliberately
+// strict: only well-formed "key = value" pairs survive, and comment prose is discarded.
+//
+// Value forms accepted:
+//   - bare tokens: lists=100, m=16, deduplicate_items=true
+//   - quoted strings: key_field='id' (pg_search bm25)
+//   - dollar-quoted blocks: options=$$ [build.internal] lists=[4096] $$ (vchord)
+
+// ExtractWithClause returns the contents of the first WITH (...) clause in s, without the
+// surrounding parentheses. Parentheses inside quoted and dollar-quoted values are ignored,
+// so a vchord TOML block survives intact. Returns "" when there is no WITH clause.
+func ExtractWithClause(s string) string {
+	lower := strings.ToLower(s)
+
+	for offset := 0; ; {
+		idx := strings.Index(lower[offset:], "with")
+		if idx < 0 {
+			return ""
+		}
+		start := offset + idx
+		offset = start + 4
+
+		// "with" must stand as its own word.
+		if start > 0 && isSQLIdentifierByte(s[start-1]) {
+			continue
+		}
+
+		pos := offset
+		for pos < len(s) && isSQLSpace(s[pos]) {
+			pos++
+		}
+		if pos >= len(s) || s[pos] != '(' {
+			continue
+		}
+
+		if end, ok := matchClosingParen(s, pos); ok {
+			return s[pos+1 : end]
+		}
+		return ""
+	}
+}
+
+// matchClosingParen returns the index of the ')' matching the '(' at open, skipping over
+// quoted and dollar-quoted spans.
+func matchClosingParen(s string, open int) (int, bool) {
+	depth := 0
+	for i := open; i < len(s); i++ {
+		switch s[i] {
+		case '\'':
+			end, ok := skipQuoted(s, i)
+			if !ok {
+				return 0, false
+			}
+			i = end
+		case '$':
+			if end, ok := skipDollarQuoted(s, i); ok {
+				i = end
+			}
+		case '(':
+			depth++
+		case ')':
+			depth--
+			if depth == 0 {
+				return i, true
+			}
+		}
+	}
+	return 0, false
+}
+
+// skipQuoted returns the index of the closing quote of the single-quoted string starting
+// at start, treating ” as an escaped quote.
+func skipQuoted(s string, start int) (int, bool) {
+	for i := start + 1; i < len(s); i++ {
+		if s[i] != '\'' {
+			continue
+		}
+		if i+1 < len(s) && s[i+1] == '\'' {
+			i++
+			continue
+		}
+		return i, true
+	}
+	return 0, false
+}
+
+// skipDollarQuoted returns the index of the last byte of the dollar-quoted block starting
+// at start ($tag$ … $tag$). Reports false when start does not open one.
+func skipDollarQuoted(s string, start int) (int, bool) {
+	tagEnd := strings.IndexByte(s[start+1:], '$')
+	if tagEnd < 0 {
+		return 0, false
+	}
+	tag := s[start : start+1+tagEnd+1]
+	for i := start + 1; i < len(tag); i++ {
+		if !isSQLIdentifierByte(tag[i]) && tag[i] != '$' {
+			return 0, false
+		}
+	}
+
+	closing := strings.Index(s[start+len(tag):], tag)
+	if closing < 0 {
+		return 0, false
+	}
+	return start + len(tag) + closing + len(tag) - 1, true
+}
+
+// SplitStorageParameters splits a WITH clause body on top-level commas, leaving quoted and
+// dollar-quoted values untouched.
+func SplitStorageParameters(clause string) []string {
+	parts := make([]string, 0, 4)
+	depth := 0
+	start := 0
+
+	for i := 0; i < len(clause); i++ {
+		switch clause[i] {
+		case '\'':
+			if end, ok := skipQuoted(clause, i); ok {
+				i = end
+			}
+		case '$':
+			if end, ok := skipDollarQuoted(clause, i); ok {
+				i = end
+			}
+		case '(', '[':
+			depth++
+		case ')', ']':
+			depth--
+		case ',':
+			if depth == 0 {
+				parts = append(parts, clause[start:i])
+				start = i + 1
+			}
+		}
+	}
+	parts = append(parts, clause[start:])
+
+	trimmed := make([]string, 0, len(parts))
+	for _, part := range parts {
+		if part = strings.TrimSpace(part); part != "" {
+			trimmed = append(trimmed, part)
+		}
+	}
+	return trimmed
+}
+
+// ParseStorageParameter splits one "key = value" storage parameter. It reports false for
+// anything that is not a well-formed parameter, which is how comment prose is filtered out.
+func ParseStorageParameter(part string) (key, value string, ok bool) {
+	key, value, found := strings.Cut(part, "=")
+	if !found {
+		return "", "", false
+	}
+
+	key = strings.ToLower(strings.TrimSpace(key))
+	value = strings.TrimSpace(value)
+	if key == "" || value == "" || !isBareIdentifier(key) {
+		return "", "", false
+	}
+	if !isStorageParameterValue(value) {
+		return "", "", false
+	}
+	return key, value, true
+}
+
+// FormatStorageParameters renders a WITH clause body as a canonical "key = value" list,
+// dropping anything malformed. Returns "" when nothing survives.
+func FormatStorageParameters(clause string) string {
+	params := make([]string, 0, 4)
+	for _, part := range SplitStorageParameters(clause) {
+		key, value, ok := ParseStorageParameter(part)
+		if !ok {
+			continue
+		}
+		params = append(params, key+" = "+value)
+	}
+	return strings.Join(params, ", ")
+}
+
+// NormalizeStorageParameterValue unquotes a value that PostgreSQL rendered as a string but
+// that is really a number, so that pg_indexes output (lists='100') and hand-written models
+// (lists=100) normalize identically. Non-numeric quoted values keep their quotes because
+// some access methods require a string (pg_search's key_field='id').
+func NormalizeStorageParameterValue(value string) string {
+	value = strings.TrimSpace(value)
+	if len(value) < 2 || value[0] != '\'' || value[len(value)-1] != '\'' {
+		return value
+	}
+
+	inner := strings.ReplaceAll(value[1:len(value)-1], "''", "'")
+	if _, err := strconv.ParseFloat(inner, 64); err == nil {
+		return inner
+	}
+	if strings.EqualFold(inner, "true") || strings.EqualFold(inner, "false") {
+		return strings.ToLower(inner)
+	}
+	return value
+}
+
+func isBareIdentifier(s string) bool {
+	if s == "" {
+		return false
+	}
+	for i := 0; i < len(s); i++ {
+		b := s[i]
+		switch {
+		case b >= 'a' && b <= 'z', b >= 'A' && b <= 'Z', b == '_':
+		case b >= '0' && b <= '9' && i > 0:
+		default:
+			return false
+		}
+	}
+	return true
+}
+
+// isStorageParameterValue reports whether value is a bare token, a complete quoted string,
+// or a complete dollar-quoted block.
+func isStorageParameterValue(value string) bool {
+	switch {
+	case value == "":
+		return false
+	case value[0] == '\'':
+		end, ok := skipQuoted(value, 0)
+		return ok && end == len(value)-1
+	case value[0] == '$':
+		end, ok := skipDollarQuoted(value, 0)
+		return ok && end == len(value)-1
+	}
+
+	for i := 0; i < len(value); i++ {
+		b := value[i]
+		switch {
+		case b >= 'a' && b <= 'z', b >= 'A' && b <= 'Z', b >= '0' && b <= '9':
+		case b == '_', b == '.', b == '-', b == '+':
+		default:
+			return false
+		}
+	}
+	return true
+}

pkg/pgsql/index_options_test.go

@@ -0,0 +1,111 @@
+package pgsql
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestExtractWithClause(t *testing.T) {
+	tests := []struct {
+		name  string
+		input string
+		want  string
+	}{
+		{"empty", "", ""},
+		{"no clause", "opclass=vector_cosine_ops", ""},
+		{"simple", "WITH (lists=100)", "lists=100"},
+		{"lowercase", "with (m = 16, ef_construction = 64)", "m = 16, ef_construction = 64"},
+		{
+			"index definition",
+			"CREATE INDEX i ON t USING ivfflat (embedding vector_cosine_ops) WITH (lists='100')",
+			"lists='100'",
+		},
+		{"paren inside quotes", "with (key_field='id(x)')", "key_field='id(x)'"},
+		{"dollar quoted", "with (options = $$f(x)$$)", "options = $$f(x)$$"},
+		{"word boundary", "swith (lists=100)", ""},
+		{"not followed by paren", "with lists=100", ""},
+		{"unterminated", "with (lists=100", ""},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := ExtractWithClause(tt.input); got != tt.want {
+				t.Errorf("ExtractWithClause(%q) = %q, want %q", tt.input, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestSplitStorageParameters(t *testing.T) {
+	tests := []struct {
+		name  string
+		input string
+		want  []string
+	}{
+		{"empty", "", []string{}},
+		{"single", "lists=100", []string{"lists=100"}},
+		{"multiple", "m = 16, ef_construction = 64", []string{"m = 16", "ef_construction = 64"}},
+		{"comma in quotes", "key_field='a,b', m=16", []string{"key_field='a,b'", "m=16"}},
+		{"comma in dollar quotes", "options=$$a,b$$, m=16", []string{"options=$$a,b$$", "m=16"}},
+		{"comma in brackets", "options=[1,2], m=16", []string{"options=[1,2]", "m=16"}},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := SplitStorageParameters(tt.input); !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("SplitStorageParameters(%q) = %v, want %v", tt.input, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestParseStorageParameter(t *testing.T) {
+	tests := []struct {
+		name      string
+		input     string
+		wantKey   string
+		wantValue string
+		wantOK    bool
+	}{
+		{"bare", "lists=100", "lists", "100", true},
+		{"spaced and uppercased key", " Lists = 100 ", "lists", "100", true},
+		{"quoted", "key_field='id'", "key_field", "'id'", true},
+		{"dollar quoted", "options=$$a$$", "options", "$$a$$", true},
+		{"boolean", "deduplicate_items=true", "deduplicate_items", "true", true},
+		{"float", "fillfactor=90.5", "fillfactor", "90.5", true},
+		{"no equals", "please drop everything", "", "", false},
+		{"empty value", "lists=", "", "", false},
+		{"quoted key rejected", "'lists'=100", "", "", false},
+		{"injection rejected", "lists=100); drop table t", "", "", false},
+		{"unterminated quote rejected", "key_field='id", "", "", false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			key, value, ok := ParseStorageParameter(tt.input)
+			if key != tt.wantKey || value != tt.wantValue || ok != tt.wantOK {
+				t.Errorf("ParseStorageParameter(%q) = (%q, %q, %v), want (%q, %q, %v)",
+					tt.input, key, value, ok, tt.wantKey, tt.wantValue, tt.wantOK)
+			}
+		})
+	}
+}
+
+func TestNormalizeStorageParameterValue(t *testing.T) {
+	tests := map[string]string{
+		"'100'":   "100",
+		"'90.5'":  "90.5",
+		"'true'":  "true",
+		"'id'":    "'id'",
+		"100":     "100",
+		"$$a,b$$": "$$a,b$$",
+		"'":       "'",
+		"''":      "''",
+	}
+
+	for input, want := range tests {
+		if got := NormalizeStorageParameterValue(input); got != want {
+			t.Errorf("NormalizeStorageParameterValue(%q) = %q, want %q", input, got, want)
+		}
+	}
+}

pkg/pgsql/types_registry.go

@@ -2,6 +2,7 @@ package pgsql
 
 import (
 	"sort"
+	"strconv"
 	"strings"
 )
 
@@ -9,6 +10,14 @@ import (
 type TypeSpec struct {
 	SupportsLength    bool
 	SupportsPrecision bool
+
+	// SupportsTypeModifier marks types whose "(...)" modifier is opaque and must be
+	// preserved verbatim (e.g. vector(1536), geometry(Point,4326)) instead of being
+	// decomposed into Length/Precision/Scale.
+	SupportsTypeModifier bool
+
+	// Extension is the PostgreSQL extension providing the type; empty for built-ins.
+	Extension string
 }
 
 var postgresBaseTypes = map[string]TypeSpec{
@@ -104,14 +113,28 @@ var postgresBaseTypes = map[string]TypeSpec{
 	"void":   {},
 
 	// Common extensions
-	"citext":    {},
-	"hstore":    {},
-	"ltree":     {},
-	"lquery":    {},
-	"ltxtquery": {},
-	"vector":    {}, // pgvector: keep explicit modifier form (vector(dim))
-	"halfvec":   {}, // pgvector: keep explicit modifier form (halfvec(dim))
-	"sparsevec": {}, // pgvector: keep explicit modifier form (sparsevec(dim))
+	"citext":    {Extension: "citext"},
+	"hstore":    {Extension: "hstore"},
+	"ltree":     {Extension: "ltree"},
+	"lquery":    {Extension: "ltree"},
+	"ltxtquery": {Extension: "ltree"},
+
+	// pgvector: modifier form is opaque (vector(dim), sparsevec(dim))
+	"vector":    {SupportsTypeModifier: true, Extension: "vector"},
+	"halfvec":   {SupportsTypeModifier: true, Extension: "vector"},
+	"sparsevec": {SupportsTypeModifier: true, Extension: "vector"},
+
+	// PostGIS: geometry/geography carry an opaque modifier (geometry(PointZ,4326))
+	"geometry":      {SupportsTypeModifier: true, Extension: "postgis"},
+	"geography":     {SupportsTypeModifier: true, Extension: "postgis"},
+	"box2d":         {Extension: "postgis"},
+	"box3d":         {Extension: "postgis"},
+	"geometry_dump": {Extension: "postgis"},
+	"geomval":       {Extension: "postgis"},
+	"spheroid":      {Extension: "postgis"},
+	"valid_detail":  {Extension: "postgis"},
+	"raster":        {SupportsTypeModifier: true, Extension: "postgis_raster"},
+	"topogeometry":  {Extension: "postgis_topology"},
 }
 
 var postgresTypeAliases = map[string]string{
@@ -346,3 +369,72 @@ func stripArraySuffixes(t string) string {
 func normalizeTypeToken(t string) string {
 	return strings.Join(strings.Fields(strings.TrimSpace(t)), " ")
 }
+
+// SupportsTypeModifier reports if this SQL type carries an opaque "(...)" modifier
+// that must be preserved verbatim (e.g. vector(1536), geometry(Point,4326)).
+func SupportsTypeModifier(sqlType string) bool {
+	base := CanonicalizeBaseType(ExtractBaseTypeLower(sqlType))
+	spec, ok := postgresBaseTypes[base]
+	return ok && spec.SupportsTypeModifier
+}
+
+// TypeExtension returns the PostgreSQL extension providing the given type
+// ("postgis", "vector", "citext", …). Built-in types return "".
+func TypeExtension(sqlType string) string {
+	base := CanonicalizeBaseType(ExtractBaseTypeLower(sqlType))
+	return postgresBaseTypes[base].Extension
+}
+
+// IsSpatialType reports whether the type comes from PostGIS (geometry, geography,
+// raster, topogeometry, …).
+func IsSpatialType(sqlType string) bool {
+	return strings.HasPrefix(TypeExtension(sqlType), "postgis")
+}
+
+// IsVectorType reports whether the type comes from pgvector (vector, halfvec, sparsevec).
+func IsVectorType(sqlType string) bool {
+	return TypeExtension(sqlType) == "vector"
+}
+
+// TypeModifier returns the raw "(...)" modifier of a SQL type without the parentheses,
+// or "" when the type has none. Array suffixes are ignored.
+// Example: geometry(PointZ,4326)[] -> "PointZ,4326".
+func TypeModifier(sqlType string) string {
+	t := stripArraySuffixes(normalizeTypeToken(sqlType))
+	start := strings.Index(t, "(")
+	end := strings.LastIndex(t, ")")
+	if start < 0 || end < start {
+		return ""
+	}
+	return strings.TrimSpace(t[start+1 : end])
+}
+
+// SpatialSRID returns the SRID declared in a PostGIS type modifier, or 0 when absent.
+// Example: geometry(Point,4326) -> 4326.
+func SpatialSRID(sqlType string) int {
+	if !IsSpatialType(sqlType) {
+		return 0
+	}
+	parts := strings.Split(TypeModifier(sqlType), ",")
+	if len(parts) < 2 {
+		return 0
+	}
+	srid, err := strconv.Atoi(strings.TrimSpace(parts[len(parts)-1]))
+	if err != nil {
+		return 0
+	}
+	return srid
+}
+
+// SpatialGeometryType returns the geometry subtype declared in a PostGIS type modifier
+// ("Point", "MultiPolygonZ", …), or "" when absent.
+func SpatialGeometryType(sqlType string) string {
+	if !IsSpatialType(sqlType) {
+		return ""
+	}
+	modifier := TypeModifier(sqlType)
+	if modifier == "" {
+		return ""
+	}
+	return strings.TrimSpace(strings.Split(modifier, ",")[0])
+}

pkg/pgsql/types_registry_test.go

@@ -145,3 +145,104 @@ func TestEquivalentSQLTypeVariants(t *testing.T) {
 		})
 	}
 }
+
+func TestExtensionTypes(t *testing.T) {
+	tests := []struct {
+		name          string
+		sqlType       string
+		wantKnown     bool
+		wantExtension string
+		wantSpatial   bool
+		wantVector    bool
+		wantModifier  bool
+	}{
+		{"geometry with modifier", "geometry(Point,4326)", true, "postgis", true, false, true},
+		{"geography", "geography", true, "postgis", true, false, true},
+		{"geometry array", "geometry[]", true, "postgis", true, false, true},
+		{"box2d", "box2d", true, "postgis", true, false, false},
+		{"raster", "raster", true, "postgis_raster", true, false, true},
+		{"topogeometry", "topogeometry", true, "postgis_topology", true, false, false},
+		{"vector", "vector(1536)", true, "vector", false, true, true},
+		{"halfvec", "halfvec(768)", true, "vector", false, true, true},
+		{"sparsevec", "sparsevec(1000)", true, "vector", false, true, true},
+		{"citext", "citext", true, "citext", false, false, false},
+		{"builtin text", "text", true, "", false, false, false},
+		{"builtin point is not postgis", "point", true, "", false, false, false},
+		{"unknown type", "mytype", false, "", false, false, false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := IsKnownPostgresType(tt.sqlType); got != tt.wantKnown {
+				t.Errorf("IsKnownPostgresType(%q) = %v, want %v", tt.sqlType, got, tt.wantKnown)
+			}
+			if got := TypeExtension(tt.sqlType); got != tt.wantExtension {
+				t.Errorf("TypeExtension(%q) = %q, want %q", tt.sqlType, got, tt.wantExtension)
+			}
+			if got := IsSpatialType(tt.sqlType); got != tt.wantSpatial {
+				t.Errorf("IsSpatialType(%q) = %v, want %v", tt.sqlType, got, tt.wantSpatial)
+			}
+			if got := IsVectorType(tt.sqlType); got != tt.wantVector {
+				t.Errorf("IsVectorType(%q) = %v, want %v", tt.sqlType, got, tt.wantVector)
+			}
+			if got := SupportsTypeModifier(tt.sqlType); got != tt.wantModifier {
+				t.Errorf("SupportsTypeModifier(%q) = %v, want %v", tt.sqlType, got, tt.wantModifier)
+			}
+		})
+	}
+}
+
+func TestExtensionTypesDoNotSupportLengthOrPrecision(t *testing.T) {
+	for _, sqlType := range []string{"geometry(Point,4326)", "geography", "vector(1536)", "halfvec(768)"} {
+		if SupportsLength(sqlType) {
+			t.Errorf("SupportsLength(%q) = true, want false", sqlType)
+		}
+		if SupportsPrecision(sqlType) {
+			t.Errorf("SupportsPrecision(%q) = true, want false", sqlType)
+		}
+	}
+}
+
+func TestSpatialTypeModifier(t *testing.T) {
+	tests := []struct {
+		sqlType      string
+		wantModifier string
+		wantGeomType string
+		wantSRID     int
+	}{
+		{"geometry(Point,4326)", "Point,4326", "Point", 4326},
+		{"geometry(MultiPolygonZ, 3857)", "MultiPolygonZ, 3857", "MultiPolygonZ", 3857},
+		{"geography(Point)", "Point", "Point", 0},
+		{"geometry", "", "", 0},
+		{"geometry(Point,4326)[]", "Point,4326", "Point", 4326},
+		{"vector(1536)", "1536", "", 0},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.sqlType, func(t *testing.T) {
+			if got := TypeModifier(tt.sqlType); got != tt.wantModifier {
+				t.Errorf("TypeModifier() = %q, want %q", got, tt.wantModifier)
+			}
+			if got := SpatialGeometryType(tt.sqlType); got != tt.wantGeomType {
+				t.Errorf("SpatialGeometryType() = %q, want %q", got, tt.wantGeomType)
+			}
+			if got := SpatialSRID(tt.sqlType); got != tt.wantSRID {
+				t.Errorf("SpatialSRID() = %d, want %d", got, tt.wantSRID)
+			}
+		})
+	}
+}
+
+func TestNormalizeEquivalentSQLTypePreservesExtensionModifiers(t *testing.T) {
+	tests := map[string]string{
+		"geometry(Point,4326)": "geometry(Point,4326)",
+		"vector(1536)":         "vector(1536)",
+		"geography(Point)[]":   "geography(Point)[]",
+	}
+
+	for input, want := range tests {
+		if got := NormalizeEquivalentSQLType(input); got != want {
+			t.Errorf("NormalizeEquivalentSQLType(%q) = %q, want %q", input, got, want)
+		}
+	}
+}

pkg/readers/bun/reader.go

@@ -245,7 +245,7 @@ func (r *Reader) getReceiverType(expr ast.Expr) string {
 }
 
 // parseTableNameMethod parses a TableName() method and extracts the table and schema name
-func (r *Reader) parseTableNameMethod(funcDecl *ast.FuncDecl) (tableName string, schemaName string) {
+func (r *Reader) parseTableNameMethod(funcDecl *ast.FuncDecl) (tableName, schemaName string) {
 	if funcDecl.Body == nil {
 		return "", ""
 	}
@@ -578,7 +578,7 @@ func (r *Reader) parseIndexesFromTag(table *models.Table, column *models.Column,
 }
 
 // extractTableNameFromTag extracts table and schema from bun tag
-func (r *Reader) extractTableNameFromTag(tag string) (tableName string, schemaName string) {
+func (r *Reader) extractTableNameFromTag(tag string) (tableName, schemaName string) {
 	// Extract bun tag value
 	re := regexp.MustCompile(`bun:"table:([^"]+)"`)
 	matches := re.FindStringSubmatch(tag)
@@ -712,12 +712,12 @@ func (r *Reader) parseTypeWithLength(typeStr string) (baseType string, length in
 		if pgsql.SupportsLength(rawBaseType) {
 			if _, err := fmt.Sscanf(matches[2], "%d", &length); err == nil {
 				baseType = pgsql.CanonicalizeBaseType(rawBaseType)
-				return
+				return baseType, length
 			}
 		}
 	}
 
-	return
+	return baseType, length
 }
 
 // goTypeToSQL maps Go types to SQL types

pkg/readers/dbml/README.md

@@ -93,6 +93,50 @@ Ref: posts.user_id > users.id [delete: cascade]
 - Indexes and composite indexes
 - Table notes and column notes
 - Enums
+- Dialect directives (`@postgres:` / `@sqlite:` — see below)
+
+## Dialect directives
+
+Lines of the form `@<namespace>[(<column>)]: <args>` embed database-specific
+features that plain DBML cannot express (partitioning, `WITHOUT ROWID`,
+tablespaces, index storage parameters, …). They are stored losslessly on the
+relevant object's `Metadata` and round-trip unchanged through the DBML writer;
+the PostgreSQL and SQLite writers translate the ones they understand to SQL.
+
+```dbml
+@postgres: search_path myapp
+
+Table myapp.events {
+  id bigint [pk]
+  created_at timestamp [not null]
+  @postgres(id): identity always
+  @postgres: partition by RANGE (created_at)
+  @sqlite: without rowid
+
+  indexes {
+    (created_at) [name: 'idx_events_created']
+    @postgres: with (fillfactor=90)
+  }
+}
+```
+
+| Position | Attaches to |
+|----------|-------------|
+| Before the first `Table {` | database |
+| Table body, no `(target)` | that table |
+| Table body, `(col)` target | column `col` (error if unknown) |
+| Inside `indexes { }` | the most recently listed index entry |
+
+`args` is preserved verbatim; the **key** (lowercased first token) drives
+duplicate detection. Repeated directives are kept in order; catalog "singleton"
+keys error on a second occurrence at the same location. All errors are
+line-numbered.
+
+`ReaderOptions.StrictDirectives` (CLI `--strict-directives`) turns an unknown
+namespace or key into an error instead of preserving it silently.
+
+See [`docs/DBML_DIRECTIVES.md`](../../../docs/DBML_DIRECTIVES.md) for the full
+grammar and the supported-directive matrix.
 
 ## Notes
 

pkg/readers/dbml/directives.go

@@ -0,0 +1,137 @@
+package dbml
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+)
+
+// directiveLineRegex matches a dialect directive line:
+//
+//	@postgres: partition by RANGE (created_at)
+//	@postgres(id): identity always
+//
+// Group 1 is the namespace, group 2 the optional (column) target, group 3 the
+// raw argument text (validated separately so error messages can be specific).
+var directiveLineRegex = regexp.MustCompile(`^@([^():]*)(?:\(([^()]*)\))?\s*:(.*)$`)
+
+// namespaceRegex is the grammar for a directive namespace.
+var namespaceRegex = regexp.MustCompile(`^[a-z][a-z0-9_]*$`)
+
+// parsedDirective is a directive line that has been parsed but not yet attached
+// to a model object.
+type parsedDirective struct {
+	namespace string
+	target    string // column name; "" when absent
+	args      string
+	line      int
+}
+
+// parseDirectiveLine parses a single "@namespace[(target)]: args" line.
+func parseDirectiveLine(line string, lineNo int) (parsedDirective, error) {
+	m := directiveLineRegex.FindStringSubmatch(line)
+	if m == nil {
+		return parsedDirective{}, fmt.Errorf(
+			"dbml: line %d: malformed directive %q (expected \"@namespace: args\")", lineNo, line)
+	}
+
+	ns := strings.TrimSpace(m[1])
+	target := strings.TrimSpace(m[2])
+	args := strings.TrimSpace(m[3])
+
+	if !namespaceRegex.MatchString(ns) {
+		return parsedDirective{}, fmt.Errorf(
+			"dbml: line %d: invalid directive namespace %q (must match [a-z][a-z0-9_]*)", lineNo, ns)
+	}
+	if args == "" {
+		return parsedDirective{}, fmt.Errorf("dbml: line %d: directive @%s has no arguments", lineNo, ns)
+	}
+	if target != "" {
+		target = stripQuotes(target)
+	}
+
+	return parsedDirective{namespace: ns, target: target, args: args, line: lineNo}, nil
+}
+
+// attachDirective resolves the target model object from the current parser state
+// and stores the directive in its Metadata, enforcing location, duplicate and
+// strict-mode rules.
+func (r *Reader) attachDirective(
+	pd parsedDirective,
+	db *models.Database,
+	table *models.Table,
+	inTable, inIndexes bool,
+	lastIndex *models.Index,
+) error {
+	strict := r.options != nil && r.options.StrictDirectives
+	key := models.DirectiveKey(pd.args)
+
+	var meta map[string]any
+	var location string
+
+	switch {
+	case inIndexes:
+		if pd.target != "" {
+			return fmt.Errorf("dbml: line %d: directive target (%s) is not allowed inside an indexes block", pd.line, pd.target)
+		}
+		if lastIndex == nil {
+			return fmt.Errorf("dbml: line %d: directive @%s must follow an index definition", pd.line, pd.namespace)
+		}
+		if lastIndex.Metadata == nil {
+			lastIndex.Metadata = make(map[string]any)
+		}
+		meta = lastIndex.Metadata
+		location = models.DirectiveLocationIndex
+
+	case inTable && table != nil:
+		if pd.target != "" {
+			col, ok := table.Columns[pd.target]
+			if !ok {
+				return fmt.Errorf("dbml: line %d: directive target column %q not found in table %q", pd.line, pd.target, table.Name)
+			}
+			if col.Metadata == nil {
+				col.Metadata = make(map[string]any)
+			}
+			meta = col.Metadata
+			location = models.DirectiveLocationColumn
+		} else {
+			if table.Metadata == nil {
+				table.Metadata = make(map[string]any)
+			}
+			meta = table.Metadata
+			location = models.DirectiveLocationTable
+		}
+
+	default:
+		if pd.target != "" {
+			return fmt.Errorf("dbml: line %d: directive target (%s) is only valid inside a table", pd.line, pd.target)
+		}
+		if db.Metadata == nil {
+			db.Metadata = make(map[string]any)
+		}
+		meta = db.Metadata
+		location = models.DirectiveLocationDatabase
+	}
+
+	spec, documented := models.LookupDirectiveSpec(pd.namespace, key)
+
+	if strict && !documented {
+		return fmt.Errorf("dbml: line %d: unknown directive @%s: %s (strict mode)", pd.line, pd.namespace, key)
+	}
+	if documented && !models.DirectiveLocationAllowed(pd.namespace, key, location) {
+		return fmt.Errorf("dbml: line %d: directive @%s: %s is not valid at %s level", pd.line, pd.namespace, key, location)
+	}
+	if documented && spec.Singleton && models.HasDirective(meta, pd.namespace, key) {
+		return fmt.Errorf("dbml: line %d: duplicate @%s directive %q at %s level", pd.line, pd.namespace, key, location)
+	}
+
+	models.AddDirective(meta, models.Directive{
+		Namespace: pd.namespace,
+		Key:       key,
+		Args:      pd.args,
+		Line:      pd.line,
+	})
+	return nil
+}

pkg/readers/dbml/directives_test.go

@@ -0,0 +1,181 @@
+package dbml
+
+import (
+	"strings"
+	"testing"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/readers"
+)
+
+func parse(t *testing.T, strict bool, src string) (*models.Database, error) {
+	t.Helper()
+	r := NewReader(&readers.ReaderOptions{StrictDirectives: strict})
+	return r.parseDBML(src)
+}
+
+func firstTable(t *testing.T, db *models.Database) *models.Table {
+	t.Helper()
+	if len(db.Schemas) == 0 || len(db.Schemas[0].Tables) == 0 {
+		t.Fatal("no table parsed")
+	}
+	return db.Schemas[0].Tables[0]
+}
+
+func TestDirectives_AttachAtEachLocation(t *testing.T) {
+	src := `@postgres: search_path myapp
+
+Table myapp.events {
+  id bigint [pk]
+  created_at timestamp [not null]
+  @postgres(id): identity always
+  @postgres: partition by RANGE (created_at)
+
+  indexes {
+    (created_at) [name: 'idx_events_created']
+    @postgres: with (fillfactor=90)
+  }
+}
+`
+	db, err := parse(t, false, src)
+	if err != nil {
+		t.Fatalf("parse: %v", err)
+	}
+
+	if !models.HasDirective(db.Metadata, "postgres", "search_path") {
+		t.Errorf("database-level directive missing: %+v", db.Metadata)
+	}
+
+	tbl := firstTable(t, db)
+	if !models.HasDirective(tbl.Metadata, "postgres", "partition") {
+		t.Errorf("table-level directive missing: %+v", tbl.Metadata)
+	}
+
+	col := tbl.Columns["id"]
+	if col == nil || !models.HasDirective(col.Metadata, "postgres", "identity") {
+		t.Errorf("column-level directive missing")
+	}
+	// Verbatim args preserved.
+	if d := models.DirectivesForNamespace(col.Metadata, "postgres"); len(d) != 1 || d[0].Args != "identity always" {
+		t.Errorf("column directive args = %+v", d)
+	}
+
+	var idx *models.Index
+	for _, i := range tbl.Indexes {
+		idx = i
+	}
+	if idx == nil || !models.HasDirective(idx.Metadata, "postgres", "with") {
+		t.Errorf("index-level directive missing: %+v", idx)
+	}
+}
+
+func TestDirectives_RepeatablePreservedAndOrdered(t *testing.T) {
+	src := `Table s.t {
+  id int [pk]
+  @postgres: with (fillfactor=90)
+  @postgres: with (autovacuum_enabled=off)
+}
+`
+	db, err := parse(t, false, src)
+	if err != nil {
+		t.Fatalf("parse: %v", err)
+	}
+	tbl := firstTable(t, db)
+	got := models.DirectivesForNamespace(tbl.Metadata, "postgres")
+	if len(got) != 2 {
+		t.Fatalf("got %d directives, want 2", len(got))
+	}
+	if got[0].Args != "with (fillfactor=90)" || got[1].Args != "with (autovacuum_enabled=off)" {
+		t.Errorf("repeatable directives out of order: %+v", got)
+	}
+}
+
+func TestDirectives_SingletonDuplicateErrors(t *testing.T) {
+	src := `Table s.t {
+  id int [pk]
+  @postgres: partition by RANGE (a)
+  @postgres: partition by LIST (b)
+}
+`
+	_, err := parse(t, false, src)
+	if err == nil || !strings.Contains(err.Error(), "duplicate") {
+		t.Fatalf("want duplicate error, got %v", err)
+	}
+	if !strings.Contains(err.Error(), "line 4") {
+		t.Errorf("error not line-numbered: %v", err)
+	}
+}
+
+func TestDirectives_MalformedErrors(t *testing.T) {
+	cases := map[string]string{
+		"no colon":       "@postgres partition by x",
+		"empty args":     "@postgres:",
+		"bad namespace":  "@Postgres: partition by x",
+		"numeric prefix": "@1x: foo",
+	}
+	for name, line := range cases {
+		t.Run(name, func(t *testing.T) {
+			src := "Table s.t {\n  id int [pk]\n  " + line + "\n}\n"
+			_, err := parse(t, false, src)
+			if err == nil {
+				t.Fatalf("want error for %q", line)
+			}
+			if !strings.Contains(err.Error(), "line 3") {
+				t.Errorf("error not line-numbered: %v", err)
+			}
+		})
+	}
+}
+
+func TestDirectives_UnknownPreservedNonStrict(t *testing.T) {
+	src := `Table s.t {
+  id int [pk]
+  @postgres: frobnicate all the things
+  @clickhouse: engine MergeTree
+}
+`
+	db, err := parse(t, false, src)
+	if err != nil {
+		t.Fatalf("parse: %v", err)
+	}
+	tbl := firstTable(t, db)
+	if !models.HasDirective(tbl.Metadata, "postgres", "frobnicate") {
+		t.Error("unknown postgres key not preserved")
+	}
+	if !models.HasDirective(tbl.Metadata, "clickhouse", "engine") {
+		t.Error("unknown namespace not preserved")
+	}
+}
+
+func TestDirectives_StrictErrors(t *testing.T) {
+	src := `Table s.t {
+  id int [pk]
+  @postgres: frobnicate x
+}
+`
+	_, err := parse(t, true, src)
+	if err == nil || !strings.Contains(err.Error(), "strict mode") {
+		t.Fatalf("want strict-mode error, got %v", err)
+	}
+}
+
+func TestDirectives_UnknownColumnTargetErrors(t *testing.T) {
+	src := `Table s.t {
+  id int [pk]
+  @postgres(missing): identity always
+}
+`
+	_, err := parse(t, false, src)
+	if err == nil || !strings.Contains(err.Error(), "not found") {
+		t.Fatalf("want unknown-column error, got %v", err)
+	}
+}
+
+func TestDirectives_WrongLocationErrors(t *testing.T) {
+	// partition is table-only.
+	src := "@postgres: partition by RANGE (x)\n\nTable s.t {\n  id int [pk]\n}\n"
+	_, err := parse(t, false, src)
+	if err == nil || !strings.Contains(err.Error(), "not valid at database level") {
+		t.Fatalf("want location error, got %v", err)
+	}
+}

pkg/readers/dbml/reader.go

@@ -434,11 +434,15 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 	var currentSchema string
 	var inIndexes bool
 	var inTable bool
+	var columnSeq uint
+	var lastIndex *models.Index // most recent index in the current Indexes block
+	lineNo := 0
 
 	tableRegex := regexp.MustCompile(`^Table\s+(.+?)\s*{`)
 	refRegex := regexp.MustCompile(`^Ref:\s+(.+)`)
 
 	for scanner.Scan() {
+		lineNo++
 		line := strings.TrimSpace(scanner.Text())
 
 		// Skip empty lines and comments
@@ -446,6 +450,20 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 			continue
 		}
 
+		// Parse a dialect directive (@postgres:, @sqlite:, …). Handled before
+		// table/column/index parsing so directive lines are never mistaken for
+		// columns.
+		if strings.HasPrefix(line, "@") {
+			pd, err := parseDirectiveLine(line, lineNo)
+			if err != nil {
+				return nil, err
+			}
+			if err := r.attachDirective(pd, db, currentTable, inTable, inIndexes, lastIndex); err != nil {
+				return nil, err
+			}
+			continue
+		}
+
 		// Parse Table definition
 		if matches := tableRegex.FindStringSubmatch(line); matches != nil {
 			tableName := matches[1]
@@ -469,11 +487,14 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 			currentTable = models.InitTable(tableName, currentSchema)
 			inTable = true
 			inIndexes = false
+			columnSeq = 0
 			continue
 		}
 
-		// End of table definition
-		if inTable && line == "}" {
+		// End of table definition. Guarded by !inIndexes so the closing brace
+		// of an `indexes { }` block is not mistaken for the end of the table
+		// (which would drop any table-level content that follows it).
+		if inTable && !inIndexes && line == "}" {
 			if currentTable != nil && currentSchema != "" {
 				schemaMap[currentSchema].Tables = append(schemaMap[currentSchema].Tables, currentTable)
 				currentTable = nil
@@ -486,20 +507,34 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 		// Parse indexes section
 		if inTable && (strings.HasPrefix(line, "Indexes {") || strings.HasPrefix(line, "indexes {")) {
 			inIndexes = true
+			lastIndex = nil
 			continue
 		}
 
 		// End of indexes section
 		if inIndexes && line == "}" {
 			inIndexes = false
+			lastIndex = nil
 			continue
 		}
 
 		// Parse index definition
 		if inIndexes && currentTable != nil {
+			// A composite `[pk]` entry inside an Indexes block declares the
+			// table's primary key (DBML's way of expressing multi-column PKs
+			// that can't be attached to a single column). It must become a
+			// primary key constraint, not a plain index, or the PK is lost.
+			if indexLineHasPKAttr(line) {
+				if constraint := r.parsePrimaryKeyIndex(line, currentTable.Name, currentSchema); constraint != nil {
+					currentTable.Constraints[constraint.Name] = constraint
+				}
+				continue
+			}
+
 			index := r.parseIndex(line, currentTable.Name, currentSchema)
 			if index != nil {
 				currentTable.Indexes[index.Name] = index
+				lastIndex = index
 			}
 			continue
 		}
@@ -516,6 +551,8 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 		if inTable && !inIndexes && currentTable != nil {
 			column, constraint := r.parseColumn(line, currentTable.Name, currentSchema)
 			if column != nil {
+				columnSeq++
+				column.Sequence = columnSeq
 				currentTable.Columns[column.Name] = column
 			}
 			if constraint != nil {
@@ -556,6 +593,28 @@ func (r *Reader) parseDBML(content string) (*models.Database, error) {
 		}
 	}
 
+	// PostgreSQL readers derive relationships from foreign keys. Do the same
+	// for DBML refs so diffing equivalent schemas compares the same model.
+	for _, schema := range schemaMap {
+		for _, table := range schema.Tables {
+			for _, constraint := range table.Constraints {
+				if constraint.Type != models.ForeignKeyConstraint {
+					continue
+				}
+				name := fmt.Sprintf("%s_to_%s", table.Name, constraint.ReferencedTable)
+				relationship := models.InitRelationship(name, models.OneToMany)
+				relationship.FromTable = table.Name
+				relationship.FromSchema = table.Schema
+				relationship.FromColumns = append([]string(nil), constraint.Columns...)
+				relationship.ToTable = constraint.ReferencedTable
+				relationship.ToSchema = constraint.ReferencedSchema
+				relationship.ToColumns = append([]string(nil), constraint.ReferencedColumns...)
+				relationship.ForeignKey = constraint.Name
+				table.Relationships[name] = relationship
+			}
+		}
+	}
+
 	// Add schemas to database
 	for _, schema := range schemaMap {
 		db.Schemas = append(db.Schemas, schema)
@@ -664,7 +723,7 @@ func (r *Reader) parseColumn(line, tableName, schemaName string) (*models.Column
 	return column, constraint
 }
 
-func splitInlineComment(line string) (content string, inlineComment string) {
+func splitInlineComment(line string) (content, inlineComment string) {
 	commentStart := strings.Index(line, "//")
 	if commentStart == -1 {
 		return line, ""
@@ -673,7 +732,7 @@ func splitInlineComment(line string) (content string, inlineComment string) {
 	return strings.TrimSpace(line[:commentStart]), strings.TrimSpace(line[commentStart+2:])
 }
 
-func splitColumnSignatureAndAttrs(line string) (signature string, attrs string) {
+func splitColumnSignatureAndAttrs(line string) (signature, attrs string) {
 	trimmed := strings.TrimSpace(line)
 	if trimmed == "" || !strings.HasSuffix(trimmed, "]") {
 		return trimmed, ""
@@ -699,7 +758,7 @@ func splitColumnSignatureAndAttrs(line string) (signature string, attrs string)
 	return trimmed, ""
 }
 
-func parseColumnSignature(signature string) (columnName string, columnType string, ok bool) {
+func parseColumnSignature(signature string) (columnName, columnType string, ok bool) {
 	signature = strings.TrimSpace(signature)
 	if signature == "" {
 		return "", "", false
@@ -743,9 +802,10 @@ func stripWrappingQuotes(s string) string {
 	return s
 }
 
-// parseIndex parses a DBML index definition
-func (r *Reader) parseIndex(line, tableName, schemaName string) *models.Index {
-	// Format: (columns) [attributes] OR columnname [attributes]
+// indexLineColumns extracts the column list from an Indexes-block entry,
+// e.g. "(col1, col2) [attrs]" or "columnname [attrs]", preserving
+// declaration order.
+func indexLineColumns(line string) []string {
 	var columns []string
 
 	// Find the attributes section to avoid parsing parentheses in notes/attributes
@@ -776,6 +836,56 @@ func (r *Reader) parseIndex(line, tableName, schemaName string) *models.Index {
 		}
 	}
 
+	return columns
+}
+
+// indexLineAttrs extracts and splits the bracketed attribute list of an
+// Indexes-block entry, e.g. "[pk]" or "[unique, name: 'foo']".
+func indexLineAttrs(line string) []string {
+	attrStart := strings.Index(line, "[")
+	attrEnd := strings.Index(line, "]")
+	if attrStart < 0 || attrEnd < 0 || attrStart >= attrEnd {
+		return nil
+	}
+
+	var attrs []string
+	for _, attr := range strings.Split(line[attrStart+1:attrEnd], ",") {
+		attrs = append(attrs, strings.TrimSpace(attr))
+	}
+	return attrs
+}
+
+// indexLineHasPKAttr reports whether an Indexes-block entry carries a `pk`
+// attribute, e.g. "(artifact_id, sha256) [pk]". DBML uses this form to
+// declare composite primary keys that can't be attached to a single column.
+func indexLineHasPKAttr(line string) bool {
+	for _, attr := range indexLineAttrs(line) {
+		if attr == "pk" || attr == "primary key" {
+			return true
+		}
+	}
+	return false
+}
+
+// parsePrimaryKeyIndex converts a composite `[pk]` entry from an Indexes
+// block into a primary key constraint, preserving the declared column order.
+func (r *Reader) parsePrimaryKeyIndex(line, tableName, schemaName string) *models.Constraint {
+	columns := indexLineColumns(line)
+	if len(columns) == 0 {
+		return nil
+	}
+
+	constraint := models.InitConstraint("pk_"+tableName, models.PrimaryKeyConstraint)
+	constraint.Schema = schemaName
+	constraint.Table = tableName
+	constraint.Columns = columns
+	return constraint
+}
+
+// parseIndex parses a DBML index definition
+func (r *Reader) parseIndex(line, tableName, schemaName string) *models.Index {
+	// Format: (columns) [attributes] OR columnname [attributes]
+	columns := indexLineColumns(line)
 	if len(columns) == 0 {
 		return nil
 	}
@@ -786,16 +896,7 @@ func (r *Reader) parseIndex(line, tableName, schemaName string) *models.Index {
 	index.Columns = columns
 
 	// Parse attributes
-	if strings.Contains(line, "[") && strings.Contains(line, "]") {
-		attrStart := strings.Index(line, "[")
-		attrEnd := strings.Index(line, "]")
-		if attrStart < attrEnd {
-			attrs := line[attrStart+1 : attrEnd]
-			attrList := strings.Split(attrs, ",")
-
-			for _, attr := range attrList {
-				attr = strings.TrimSpace(attr)
-
+	for _, attr := range indexLineAttrs(line) {
 		if attr == "unique" {
 			index.Unique = true
 		} else if strings.HasPrefix(attr, "name:") {
@@ -806,8 +907,6 @@ func (r *Reader) parseIndex(line, tableName, schemaName string) *models.Index {
 			index.Type = strings.Trim(indexType, "'\"")
 		}
 	}
-		}
-	}
 
 	// Generate name if not provided
 	if index.Name == "" {
@@ -964,5 +1063,5 @@ func (r *Reader) parseTableRef(ref string) (schema, table string, columns []stri
 		table = stripQuotes(parts[0])
 	}
 
-	return
+	return schema, table, columns
 }

pkg/readers/dbml/reader_test.go

@@ -689,7 +689,7 @@ func TestReadDirectory_CommentedRefsLast(t *testing.T) {
 func TestReadDirectory_EmptyDirectory(t *testing.T) {
 	// Create a temporary empty directory
 	tmpDir := filepath.Join("..", "..", "..", "tests", "assets", "dbml", "empty_test_dir")
-	err := os.MkdirAll(tmpDir, 0755)
+	err := os.MkdirAll(tmpDir, 0o755)
 	if err != nil {
 		t.Fatalf("Failed to create temp directory: %v", err)
 	}
@@ -863,6 +863,13 @@ func TestParseColumn_PostgresTypes(t *testing.T) {
 			wantName: "embedding",
 			wantType: "vector(1536)",
 		},
+		{
+			name:        "postgis geometry with type modifier",
+			line:        "location geometry(Point,4326) [not null]",
+			wantName:    "location",
+			wantType:    "geometry(Point,4326)",
+			wantNotNull: true,
+		},
 		{
 			name:     "multi word timestamp type",
 			line:     "published_at timestamp with time zone",
@@ -932,3 +939,97 @@ func TestHasCommentedRefs(t *testing.T) {
 		})
 	}
 }
+
+// TestReader_CompositePKIndex verifies that a composite `[pk]` entry inside
+// an Indexes block is turned into a primary key constraint, in declaration
+// order, rather than being silently dropped.
+func TestReader_CompositePKIndex(t *testing.T) {
+	dbmlContent := `Table artifact_blob {
+  artifact_id integer [not null]
+  sha256 text [not null]
+  size integer
+
+  Indexes {
+    (artifact_id, sha256) [pk]
+  }
+}
+`
+	dir := t.TempDir()
+	path := filepath.Join(dir, "composite_pk.dbml")
+	if err := os.WriteFile(path, []byte(dbmlContent), 0o644); err != nil {
+		t.Fatalf("failed to write fixture: %v", err)
+	}
+
+	reader := NewReader(&readers.ReaderOptions{FilePath: path})
+	db, err := reader.ReadDatabase()
+	if err != nil {
+		t.Fatalf("ReadDatabase() error = %v", err)
+	}
+
+	table := db.Schemas[0].Tables[0]
+
+	var pk *models.Constraint
+	for _, c := range table.Constraints {
+		if c.Type == models.PrimaryKeyConstraint {
+			pk = c
+			break
+		}
+	}
+	if pk == nil {
+		t.Fatal("expected a primary key constraint, got none")
+	}
+	want := []string{"artifact_id", "sha256"}
+	if len(pk.Columns) != len(want) {
+		t.Fatalf("expected PK columns %v, got %v", want, pk.Columns)
+	}
+	for i, col := range want {
+		if pk.Columns[i] != col {
+			t.Errorf("PK column[%d] = %q, want %q (order must match declaration)", i, pk.Columns[i], col)
+		}
+	}
+
+	// No plain index should be emitted for the pk-only entry.
+	if len(table.Indexes) != 0 {
+		t.Errorf("expected no plain indexes from a [pk] Indexes entry, got %v", table.Indexes)
+	}
+}
+
+// TestReader_ColumnPKOrderPreserved verifies that composite primary keys
+// declared via column-level [pk] attributes keep declaration order (via
+// Column.Sequence) instead of falling back to alphabetical sorting.
+func TestReader_ColumnPKOrderPreserved(t *testing.T) {
+	dbmlContent := `Table snapshot_artifact {
+  snapshot_id integer [pk, not null]
+  artifact_id integer [pk, not null]
+}
+`
+	dir := t.TempDir()
+	path := filepath.Join(dir, "column_pk_order.dbml")
+	if err := os.WriteFile(path, []byte(dbmlContent), 0o644); err != nil {
+		t.Fatalf("failed to write fixture: %v", err)
+	}
+
+	reader := NewReader(&readers.ReaderOptions{FilePath: path})
+	db, err := reader.ReadDatabase()
+	if err != nil {
+		t.Fatalf("ReadDatabase() error = %v", err)
+	}
+
+	table := db.Schemas[0].Tables[0]
+
+	snapshotCol, ok := table.Columns["snapshot_id"]
+	if !ok {
+		t.Fatal("column 'snapshot_id' not found")
+	}
+	artifactCol, ok := table.Columns["artifact_id"]
+	if !ok {
+		t.Fatal("column 'artifact_id' not found")
+	}
+
+	if snapshotCol.Sequence == 0 || artifactCol.Sequence == 0 {
+		t.Fatalf("expected non-zero Sequence values, got snapshot_id=%d artifact_id=%d", snapshotCol.Sequence, artifactCol.Sequence)
+	}
+	if snapshotCol.Sequence >= artifactCol.Sequence {
+		t.Errorf("expected snapshot_id (declared first) to have a lower Sequence than artifact_id, got %d >= %d", snapshotCol.Sequence, artifactCol.Sequence)
+	}
+}

pkg/readers/gorm/reader.go

@@ -246,7 +246,7 @@ func (r *Reader) getReceiverType(expr ast.Expr) string {
 }
 
 // parseTableNameMethod parses a TableName() method and extracts the table and schema name
-func (r *Reader) parseTableNameMethod(funcDecl *ast.FuncDecl) (tableName string, schemaName string) {
+func (r *Reader) parseTableNameMethod(funcDecl *ast.FuncDecl) (tableName, schemaName string) {
 	if funcDecl.Body == nil {
 		return "", ""
 	}
@@ -669,7 +669,7 @@ func (r *Reader) parseIndexesFromTag(table *models.Table, column *models.Column,
 }
 
 // extractTableFromGormTag extracts table and schema from gorm tag
-func (r *Reader) extractTableFromGormTag(tag string) (tablename string, schemaName string) {
+func (r *Reader) extractTableFromGormTag(tag string) (tablename, schemaName string) {
 	// This is typically set via TableName() method, not in tags
 	// We'll return empty strings and rely on deriveTableName
 	return "", ""
@@ -794,12 +794,12 @@ func (r *Reader) parseTypeWithLength(typeStr string) (baseType string, length in
 		if pgsql.SupportsLength(rawBaseType) && !strings.Contains(parens, ",") {
 			if _, err := fmt.Sscanf(parens, "%d", &length); err == nil {
 				baseType = pgsql.CanonicalizeBaseType(rawBaseType)
-				return
+				return baseType, length
 			}
 		}
 	}
 
-	return
+	return baseType, length
 }
 
 // parseTypeWithReferences parses a type string and extracts base type, length, and references
@@ -816,12 +816,12 @@ func (r *Reader) parseTypeWithReferences(typeStr string) (baseType string, lengt
 
 		// Parse base type for length
 		baseType, length = r.parseTypeWithLength(baseTypePart)
-		return
+		return baseType, length, refInfo
 	}
 
 	// No references, just parse type and length
 	baseType, length = r.parseTypeWithLength(typeStr)
-	return
+	return baseType, length, refInfo
 }
 
 // parseGormTag parses a gorm tag string into a map

pkg/readers/graphql/type_mapping.go

@@ -32,7 +32,7 @@ func (r *Reader) isScalarType(typeName string, ctx *parseContext) bool {
 	return commonCustomScalars[typeName]
 }
 
-func (r *Reader) graphQLTypeToSQL(gqlType string, fieldName string, typeName string) string {
+func (r *Reader) graphQLTypeToSQL(gqlType, fieldName, typeName string) string {
 	// Check for ID type with configurable mapping
 	if gqlType == "ID" {
 		// Check metadata for ID type preference

pkg/readers/mssql/reader_test.go

@@ -3,9 +3,10 @@ package mssql
 import (
 	"testing"
 
+	"github.com/stretchr/testify/assert"
+
 	"git.warky.dev/wdevs/relspecgo/pkg/mssql"
 	"git.warky.dev/wdevs/relspecgo/pkg/readers"
-	"github.com/stretchr/testify/assert"
 )
 
 // TestMapDataType tests MSSQL type mapping to canonical types

pkg/readers/pgsql/README.md

@@ -128,6 +128,27 @@ sessions so they are identifiable in `pg_stat_activity`. If you provide
 - Sequence properties
 - Associated tables
 
+## Extension Types (PostGIS, pgvector)
+
+- Extension column types keep their catalog-formatted form: `geometry(Point,4326)`,
+  `geography(Point)`, `vector(1536)`, `halfvec(768)`, `citext`, arrays included.
+- Built-in types are canonicalized and their dimensions moved to
+  `Column.Length` / `Precision` / `Scale`; extension modifiers stay in `Column.Type`.
+- Index access methods are read from the definition as-is: `gist`, `spgist`, `brin`, `hnsw`,
+  `ivfflat`, `vchordrq`, `vchordg`, `bm25`.
+- Operator class and `WITH (...)` parameters have no model field, so they are stored in
+  `Index.Comment` in the form the PostgreSQL writer reads back:
+
+  ```
+  opclass=vector_cosine_ops; with (m=16, ef_construction=64)
+  ```
+
+  Ordering modifiers (`DESC`, `NULLS LAST`, `COLLATE`) are not treated as operator classes.
+  Numeric parameter values are unquoted (`lists='100'` -> `lists=100`); string values keep
+  their quotes (`key_field='id'`), and dollar-quoted values are preserved whole.
+- Installed extensions are read from `pg_extension` into `schema.Metadata["extensions"]`
+  (only extensions RelSpec recognizes), so a read/write round-trip re-creates them.
+
 ## Notes
 
 - Requires PostgreSQL connection permissions

pkg/readers/pgsql/queries.go

@@ -5,6 +5,7 @@ import (
 	"strings"
 
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/pgsql"
 )
 
 // querySchemas retrieves all non-system schemas from the database
@@ -46,6 +47,41 @@ func (r *Reader) querySchemas() ([]*models.Schema, error) {
 	return schemas, rows.Err()
 }
 
+// queryExtensions retrieves the extensions installed into a schema. Only extensions RelSpec
+// recognizes are kept, so a round-trip never emits a CREATE EXTENSION the writer cannot
+// order; plpgsql is not registered and is therefore skipped along with other built-ins.
+func (r *Reader) queryExtensions(schemaName string) ([]string, error) {
+	query := `
+		SELECT e.extname
+		FROM pg_extension e
+		JOIN pg_namespace n ON n.oid = e.extnamespace
+		WHERE n.nspname = $1
+		ORDER BY e.extname
+	`
+
+	rows, err := r.conn.Query(r.ctx, query, schemaName)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	extensions := make([]string, 0)
+	for rows.Next() {
+		var name string
+		if err := rows.Scan(&name); err != nil {
+			return nil, err
+		}
+		if pgsql.IsKnownExtension(name) {
+			extensions = append(extensions, name)
+		}
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+
+	return pgsql.SortExtensions(extensions), nil
+}
+
 // queryTables retrieves all tables for a given schema
 func (r *Reader) queryTables(schemaName string) ([]*models.Table, error) {
 	query := `
@@ -502,8 +538,13 @@ func (r *Reader) queryCheckConstraints(schemaName string) (map[string][]*models.
 		FROM information_schema.table_constraints tc
 		JOIN information_schema.check_constraints cc
 			ON tc.constraint_name = cc.constraint_name
+			AND cc.constraint_schema = tc.table_schema
+		JOIN pg_catalog.pg_constraint pc
+			ON pc.conname = tc.constraint_name
+			AND pc.connamespace = (SELECT oid FROM pg_namespace WHERE nspname = tc.table_schema)
 		WHERE tc.constraint_type = 'CHECK'
 		  AND tc.table_schema = $1
+		AND pc.contype = 'c'
 	`
 
 	rows, err := r.conn.Query(r.ctx, query, schemaName)
@@ -543,7 +584,12 @@ func (r *Reader) queryIndexes(schemaName string) (map[string][]*models.Index, er
 			indexname,
 			indexdef
 		FROM pg_indexes
+		JOIN pg_catalog.pg_class idx ON idx.relname = indexname
+		JOIN pg_catalog.pg_index i ON i.indexrelid = idx.oid
+		JOIN pg_catalog.pg_namespace idx_ns ON idx_ns.oid = idx.relnamespace
 		WHERE schemaname = $1
+		  AND idx_ns.nspname = schemaname
+		  AND NOT i.indisprimary
 		ORDER BY schemaname, tablename, indexname
 	`
 
@@ -597,6 +643,7 @@ func (r *Reader) parseIndexDefinition(indexName, tableName, schema, indexDef str
 	}
 
 	// Extract columns - pattern: (column1, column2, ...)
+	opClass := ""
 	columnsRegex := regexp.MustCompile(`\(([^)]+)\)`)
 	if matches := columnsRegex.FindStringSubmatch(indexDef); len(matches) > 1 {
 		columnsStr := matches[1]
@@ -604,8 +651,17 @@ func (r *Reader) parseIndexDefinition(indexName, tableName, schema, indexDef str
 		columnParts := strings.Split(columnsStr, ",")
 		for _, col := range columnParts {
 			col = strings.TrimSpace(col)
+			fields := strings.Fields(col)
+			if len(fields) == 0 {
+				continue
+			}
+			// Remember an explicit operator class (e.g. "embedding vector_cosine_ops")
+			// so the writer can reproduce it; ordering modifiers are not operator classes.
+			if opClass == "" && len(fields) > 1 {
+				opClass = extractIndexOperatorClass(fields[1:])
+			}
 			// Remove any ordering (ASC/DESC) or other modifiers
-			col = strings.Fields(col)[0]
+			col = fields[0]
 			// Remove parentheses if it's an expression
 			if !strings.Contains(col, "(") {
 				index.Columns = append(index.Columns, col)
@@ -613,6 +669,15 @@ func (r *Reader) parseIndexDefinition(indexName, tableName, schema, indexDef str
 		}
 	}
 
+	// Extract access method storage parameters, e.g. WITH (lists='100')
+	storageParams := normalizeIndexStorageParams(pgsql.ExtractWithClause(indexDef))
+
+	// Operator class and storage parameters have no dedicated model fields; carry them in
+	// the comment hint the PostgreSQL writer reads back.
+	if hint := buildIndexHint(opClass, storageParams); hint != "" && index.Comment == "" {
+		index.Comment = hint
+	}
+
 	// Extract WHERE clause for partial indexes
 	whereRegex := regexp.MustCompile(`WHERE\s+(.+)$`)
 	if matches := whereRegex.FindStringSubmatch(indexDef); len(matches) > 1 {
@@ -622,6 +687,52 @@ func (r *Reader) parseIndexDefinition(indexName, tableName, schema, indexDef str
 	return index, nil
 }
 
+// indexOrderingKeywords are column modifiers that are not operator classes.
+var indexOrderingKeywords = map[string]bool{
+	"asc": true, "desc": true, "nulls": true, "first": true, "last": true, "collate": true,
+}
+
+// extractIndexOperatorClass picks the operator class out of a column's trailing modifiers.
+// Returns "" when the modifiers are only ordering keywords.
+func extractIndexOperatorClass(modifiers []string) string {
+	for _, modifier := range modifiers {
+		lower := strings.ToLower(strings.TrimSpace(modifier))
+		if lower == "" || indexOrderingKeywords[lower] {
+			continue
+		}
+		return lower
+	}
+	return ""
+}
+
+// normalizeIndexStorageParams rewrites "m='16', ef_construction='64'" as "m=16,
+// ef_construction=64". Non-numeric values keep their quotes because some access methods
+// require a string literal (pg_search's key_field='id').
+func normalizeIndexStorageParams(params string) string {
+	normalized := make([]string, 0, 4)
+	for _, part := range pgsql.SplitStorageParameters(params) {
+		key, value, ok := pgsql.ParseStorageParameter(part)
+		if !ok {
+			continue
+		}
+		normalized = append(normalized, key+"="+pgsql.NormalizeStorageParameterValue(value))
+	}
+	return strings.Join(normalized, ", ")
+}
+
+// buildIndexHint renders the operator class and storage parameters in the form the
+// PostgreSQL writer parses back out of an index comment.
+func buildIndexHint(opClass, storageParams string) string {
+	parts := make([]string, 0, 2)
+	if opClass != "" {
+		parts = append(parts, "opclass="+opClass)
+	}
+	if storageParams != "" {
+		parts = append(parts, "with ("+storageParams+")")
+	}
+	return strings.Join(parts, "; ")
+}
+
 // normalizePostgresDefault converts a raw PostgreSQL column_default expression into the
 // unquoted string value that the model convention expects. PostgreSQL stores string
 // literal defaults as 'value' or 'value'::type (e.g. '{}'::text[]), while every other

pkg/readers/pgsql/reader.go

@@ -88,6 +88,18 @@ func (r *Reader) ReadDatabase() (*models.Database, error) {
 		}
 		schema.Sequences = sequences
 
+		// Query extensions installed into this schema
+		extensions, err := r.queryExtensions(schema.Name)
+		if err != nil {
+			return nil, fmt.Errorf("failed to query extensions for schema %s: %w", schema.Name, err)
+		}
+		if len(extensions) > 0 {
+			if schema.Metadata == nil {
+				schema.Metadata = make(map[string]any)
+			}
+			schema.Metadata["extensions"] = extensions
+		}
+
 		// Query columns for tables and views
 		columnsMap, err := r.queryColumns(schema.Name)
 		if err != nil {
@@ -278,11 +290,6 @@ func (r *Reader) mapDataType(pgType, udtName, formattedType string, hasNextval b
 		}
 	}
 
-	// information_schema reports arrays generically as "ARRAY" with udt_name like "_text".
-	if strings.EqualFold(pgType, "ARRAY") && strings.HasPrefix(udtName, "_") && len(udtName) > 1 {
-		return udtName[1:] + "[]"
-	}
-
 	// Use the database-formatted type when available. For known built-in types, strip
 	// embedded dimensions (they are stored in column.Length/Precision/Scale separately).
 	// For unknown/custom types, keep the full formatted string (e.g. vector(1536)).
@@ -303,6 +310,13 @@ func (r *Reader) mapDataType(pgType, udtName, formattedType string, hasNextval b
 		return formattedType
 	}
 
+	// information_schema reports arrays generically as "ARRAY" with udt_name like "_text".
+	// Only reached when the catalog-formatted type is unavailable, which is the one case
+	// where the element modifier (e.g. geometry(Point,4326)[]) cannot be recovered.
+	if strings.EqualFold(pgType, "ARRAY") && strings.HasPrefix(udtName, "_") && len(udtName) > 1 {
+		return udtName[1:] + "[]"
+	}
+
 	// Fall back to normalizing the information_schema type name directly.
 	canonical := pgsql.NormalizePGType(normalizedPGType)
 	if pgsql.IsKnownPGBaseType(canonical) {
@@ -327,8 +341,10 @@ func (r *Reader) deriveRelationship(table *models.Table, fk *models.Constraint)
 	relationship := models.InitRelationship(relationshipName, models.OneToMany)
 	relationship.FromTable = table.Name
 	relationship.FromSchema = table.Schema
+	relationship.FromColumns = append([]string(nil), fk.Columns...)
 	relationship.ToTable = fk.ReferencedTable
 	relationship.ToSchema = fk.ReferencedSchema
+	relationship.ToColumns = append([]string(nil), fk.ReferencedColumns...)
 	relationship.ForeignKey = fk.Name
 
 	// Store constraint actions in properties

pkg/readers/pgsql/reader_test.go

@@ -2,6 +2,7 @@ package pgsql
 
 import (
 	"os"
+	"reflect"
 	"testing"
 
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
@@ -359,6 +360,14 @@ func TestDeriveRelationship(t *testing.T) {
 		t.Errorf("Expected ToTable 'users', got '%s'", rel.ToTable)
 	}
 
+	if !reflect.DeepEqual(rel.FromColumns, []string{"user_id"}) {
+		t.Errorf("Expected FromColumns [user_id], got %v", rel.FromColumns)
+	}
+
+	if !reflect.DeepEqual(rel.ToColumns, []string{"id"}) {
+		t.Errorf("Expected ToColumns [id], got %v", rel.ToColumns)
+	}
+
 	if rel.ForeignKey != "fk_orders_user_id" {
 		t.Errorf("Expected ForeignKey 'fk_orders_user_id', got '%s'", rel.ForeignKey)
 	}
@@ -392,3 +401,101 @@ func BenchmarkReader_ReadDatabase(b *testing.B) {
 		}
 	}
 }
+
+func TestParseIndexDefinition_ExtensionIndexes(t *testing.T) {
+	reader := &Reader{}
+
+	tests := []struct {
+		name        string
+		indexDef    string
+		wantType    string
+		wantColumns []string
+		wantComment string
+	}{
+		{
+			name:        "hnsw vector index with storage parameters",
+			indexDef:    "CREATE INDEX idx_docs_embedding ON public.docs USING hnsw (embedding vector_cosine_ops) WITH (m='16', ef_construction='64')",
+			wantType:    "hnsw",
+			wantColumns: []string{"embedding"},
+			wantComment: "opclass=vector_cosine_ops; with (m=16, ef_construction=64)",
+		},
+		{
+			name:        "ivfflat vector index",
+			indexDef:    "CREATE INDEX idx_docs_embedding ON public.docs USING ivfflat (embedding vector_l2_ops) WITH (lists='100')",
+			wantType:    "ivfflat",
+			wantColumns: []string{"embedding"},
+			wantComment: "opclass=vector_l2_ops; with (lists=100)",
+		},
+		{
+			name:        "gist geometry index with default operator class",
+			indexDef:    "CREATE INDEX idx_places_geom ON public.places USING gist (geom)",
+			wantType:    "gist",
+			wantColumns: []string{"geom"},
+			wantComment: "",
+		},
+		{
+			name:        "gist geometry index with explicit operator class",
+			indexDef:    "CREATE INDEX idx_places_geom ON public.places USING gist (geom gist_geometry_ops_nd)",
+			wantType:    "gist",
+			wantColumns: []string{"geom"},
+			wantComment: "opclass=gist_geometry_ops_nd",
+		},
+		{
+			name:        "btree ordering modifiers are not operator classes",
+			indexDef:    "CREATE INDEX idx_users_created ON public.users USING btree (created_at DESC NULLS LAST)",
+			wantType:    "btree",
+			wantColumns: []string{"created_at"},
+			wantComment: "",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			index, err := reader.parseIndexDefinition("idx", "tbl", "public", tt.indexDef)
+			if err != nil {
+				t.Fatalf("parseIndexDefinition() error = %v", err)
+			}
+
+			if index.Type != tt.wantType {
+				t.Errorf("Type = %q, want %q", index.Type, tt.wantType)
+			}
+			if len(index.Columns) != len(tt.wantColumns) {
+				t.Fatalf("Columns = %v, want %v", index.Columns, tt.wantColumns)
+			}
+			for i, col := range tt.wantColumns {
+				if index.Columns[i] != col {
+					t.Errorf("Columns[%d] = %q, want %q", i, index.Columns[i], col)
+				}
+			}
+			if index.Comment != tt.wantComment {
+				t.Errorf("Comment = %q, want %q", index.Comment, tt.wantComment)
+			}
+		})
+	}
+}
+
+func TestMapDataType_ExtensionTypesPreserveModifiers(t *testing.T) {
+	reader := &Reader{}
+
+	tests := []struct {
+		name          string
+		pgType        string
+		udtName       string
+		formattedType string
+		want          string
+	}{
+		{"postgis geometry", "USER-DEFINED", "geometry", "geometry(Point,4326)", "geometry(Point,4326)"},
+		{"postgis geography", "USER-DEFINED", "geography", "geography(Point,4326)", "geography(Point,4326)"},
+		{"postgis geometry without modifier", "USER-DEFINED", "geometry", "geometry", "geometry"},
+		{"pgvector halfvec", "USER-DEFINED", "halfvec", "halfvec(768)", "halfvec(768)"},
+		{"postgis geometry array", "ARRAY", "_geometry", "geometry(Point,4326)[]", "geometry(Point,4326)[]"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := reader.mapDataType(tt.pgType, tt.udtName, tt.formattedType, false); got != tt.want {
+				t.Errorf("mapDataType() = %q, want %q", got, tt.want)
+			}
+		})
+	}
+}

pkg/readers/prisma/reader_test.go

@@ -28,7 +28,7 @@ model User {
   id Int @id @default(autoincrement())
 }`
 
-	if err := os.WriteFile(schemaPath, []byte(content), 0644); err != nil {
+	if err := os.WriteFile(schemaPath, []byte(content), 0o644); err != nil {
 		t.Fatalf("failed to write schema: %v", err)
 	}
 
@@ -58,7 +58,7 @@ model User {
   id Int @id @default(autoincrement())
 }`
 
-	if err := os.WriteFile(schemaPath, []byte(content), 0644); err != nil {
+	if err := os.WriteFile(schemaPath, []byte(content), 0o644); err != nil {
 		t.Fatalf("failed to write schema: %v", err)
 	}
 

pkg/readers/reader.go

@@ -28,6 +28,10 @@ type ReaderOptions struct {
 	// Prisma7 enables Prisma 7-specific handling for Prisma schemas.
 	Prisma7 bool
 
+	// StrictDirectives makes DBML dialect directives (@postgres:, @sqlite:, …)
+	// fail on an unknown namespace or key instead of preserving them silently.
+	StrictDirectives bool
+
 	// Additional options can be added here as needed
 	Metadata map[string]interface{}
 }

pkg/readers/sqldir/reader.go

@@ -175,7 +175,6 @@ func (r *Reader) readScripts() ([]*models.Script, error) {
 
 		return nil
 	})
-
 	if err != nil {
 		return nil, err
 	}

pkg/readers/sqldir/reader_test.go

@@ -30,18 +30,18 @@ func TestReader_ReadDatabase(t *testing.T) {
 
 	for filename, content := range testFiles {
 		filePath := filepath.Join(tempDir, filename)
-		if err := os.WriteFile(filePath, []byte(content), 0644); err != nil {
+		if err := os.WriteFile(filePath, []byte(content), 0o644); err != nil {
 			t.Fatalf("Failed to create test file %s: %v", filename, err)
 		}
 	}
 
 	// Create subdirectory with additional script
 	subDir := filepath.Join(tempDir, "migrations")
-	if err := os.MkdirAll(subDir, 0755); err != nil {
+	if err := os.MkdirAll(subDir, 0o755); err != nil {
 		t.Fatalf("Failed to create subdirectory: %v", err)
 	}
 	subFile := filepath.Join(subDir, "3_001_add_column.sql")
-	if err := os.WriteFile(subFile, []byte("ALTER TABLE users ADD COLUMN email TEXT;"), 0644); err != nil {
+	if err := os.WriteFile(subFile, []byte("ALTER TABLE users ADD COLUMN email TEXT;"), 0o644); err != nil {
 		t.Fatalf("Failed to create subdirectory file: %v", err)
 	}
 
@@ -141,7 +141,7 @@ func TestReader_ReadSchema(t *testing.T) {
 
 	// Create test SQL file
 	testFile := filepath.Join(tempDir, "1_001_test.sql")
-	if err := os.WriteFile(testFile, []byte("SELECT 1;"), 0644); err != nil {
+	if err := os.WriteFile(testFile, []byte("SELECT 1;"), 0o644); err != nil {
 		t.Fatalf("Failed to create test file: %v", err)
 	}
 
@@ -220,14 +220,14 @@ func TestReader_InvalidFilename(t *testing.T) {
 
 	for _, filename := range invalidFiles {
 		filePath := filepath.Join(tempDir, filename)
-		if err := os.WriteFile(filePath, []byte("SELECT 1;"), 0644); err != nil {
+		if err := os.WriteFile(filePath, []byte("SELECT 1;"), 0o644); err != nil {
 			t.Fatalf("Failed to create test file %s: %v", filename, err)
 		}
 	}
 
 	// Create one valid file
 	validFile := filepath.Join(tempDir, "1_001_valid.sql")
-	if err := os.WriteFile(validFile, []byte("SELECT 1;"), 0644); err != nil {
+	if err := os.WriteFile(validFile, []byte("SELECT 1;"), 0o644); err != nil {
 		t.Fatalf("Failed to create valid file: %v", err)
 	}
 
@@ -277,7 +277,7 @@ func TestReader_HyphenFormat(t *testing.T) {
 
 	for filename, content := range testFiles {
 		filePath := filepath.Join(tempDir, filename)
-		if err := os.WriteFile(filePath, []byte(content), 0644); err != nil {
+		if err := os.WriteFile(filePath, []byte(content), 0o644); err != nil {
 			t.Fatalf("Failed to create test file %s: %v", filename, err)
 		}
 	}
@@ -343,7 +343,7 @@ func TestReader_MixedFormat(t *testing.T) {
 
 	for filename, content := range testFiles {
 		filePath := filepath.Join(tempDir, filename)
-		if err := os.WriteFile(filePath, []byte(content), 0644); err != nil {
+		if err := os.WriteFile(filePath, []byte(content), 0o644); err != nil {
 			t.Fatalf("Failed to create test file %s: %v", filename, err)
 		}
 	}
@@ -386,13 +386,13 @@ func TestReader_SkipSymlinks(t *testing.T) {
 
 	// Create a real SQL file
 	realFile := filepath.Join(tempDir, "1_001_real_file.sql")
-	if err := os.WriteFile(realFile, []byte("SELECT 1;"), 0644); err != nil {
+	if err := os.WriteFile(realFile, []byte("SELECT 1;"), 0o644); err != nil {
 		t.Fatalf("Failed to create real file: %v", err)
 	}
 
 	// Create another file to link to
 	targetFile := filepath.Join(tempDir, "2_001_target.sql")
-	if err := os.WriteFile(targetFile, []byte("SELECT 2;"), 0644); err != nil {
+	if err := os.WriteFile(targetFile, []byte("SELECT 2;"), 0o644); err != nil {
 		t.Fatalf("Failed to create target file: %v", err)
 	}
 

pkg/reflectutil/helpers.go

@@ -192,7 +192,7 @@ func mapKeyLess(a, b reflect.Value) bool {
 
 // MapGet safely gets a value from a map by key
 // Returns nil if key doesn't exist or not a map
-func MapGet(m interface{}, key interface{}) interface{} {
+func MapGet(m, key interface{}) interface{} {
 	v := reflect.ValueOf(m)
 	v, ok := Deref(v)
 	if !ok {

pkg/sqltypes/sql_types.go

@@ -111,6 +111,7 @@ func (n *SqlNull[T]) Scan(value any) error {
 		return n.FromString(fmt.Sprintf("%v", value))
 	}
 }
+
 func (n *SqlNull[T]) FromString(s string) error {
 	s = strings.TrimSpace(s)
 	n.Valid = false
@@ -444,55 +445,55 @@ type (
 	SqlUUID      = SqlNull[uuid.UUID]
 )
 
-// SqlTimeStamp - Timestamp with custom formatting (YYYY-MM-DDTHH:MM:SS).
+// SqlTimeStamp - Timestamp serialized as RFC3339.
 type SqlTimeStamp struct{ SqlNull[time.Time] }
 
 func (t SqlTimeStamp) MarshalJSON() ([]byte, error) {
-	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0002, 1, 1, 0, 0, 0, 0, time.UTC)) {
+	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0o002, 1, 1, 0, 0, 0, 0, time.UTC)) {
 		return []byte("null"), nil
 	}
-	return fmt.Appendf(nil, `"%s"`, t.Val.Format("2006-01-02T15:04:05")), nil
+	return fmt.Appendf(nil, `"%s"`, t.Val.Format(time.RFC3339)), nil
 }
 
 func (t *SqlTimeStamp) UnmarshalJSON(b []byte) error {
 	if err := t.SqlNull.UnmarshalJSON(b); err != nil {
 		return err
 	}
-	if t.Valid && (t.Val.IsZero() || t.Val.Format("2006-01-02T15:04:05") == "0001-01-01T00:00:00") {
+	if t.Valid && (t.Val.IsZero() || t.Val.Format(time.RFC3339) == "0001-01-01T00:00:00Z") {
 		t.Valid = false
 	}
 	return nil
 }
 
 func (t SqlTimeStamp) Value() (driver.Value, error) {
-	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0002, 1, 1, 0, 0, 0, 0, time.UTC)) {
+	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0o002, 1, 1, 0, 0, 0, 0, time.UTC)) {
 		return nil, nil
 	}
-	return t.Val.Format("2006-01-02T15:04:05"), nil
+	return t.Val.Format(time.RFC3339), nil
 }
 
 func (t SqlTimeStamp) MarshalYAML() (any, error) {
-	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0002, 1, 1, 0, 0, 0, 0, time.UTC)) {
+	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0o002, 1, 1, 0, 0, 0, 0, time.UTC)) {
 		return nil, nil
 	}
-	return t.Val.Format("2006-01-02T15:04:05"), nil
+	return t.Val.Format(time.RFC3339), nil
 }
 
 func (t *SqlTimeStamp) UnmarshalYAML(value *yaml.Node) error {
 	if err := t.SqlNull.UnmarshalYAML(value); err != nil {
 		return err
 	}
-	if t.Valid && (t.Val.IsZero() || t.Val.Format("2006-01-02T15:04:05") == "0001-01-01T00:00:00") {
+	if t.Valid && (t.Val.IsZero() || t.Val.Format(time.RFC3339) == "0001-01-01T00:00:00Z") {
 		t.Valid = false
 	}
 	return nil
 }
 
 func (t SqlTimeStamp) MarshalXML(e *xml.Encoder, start xml.StartElement) error {
-	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0002, 1, 1, 0, 0, 0, 0, time.UTC)) {
+	if !t.Valid || t.Val.IsZero() || t.Val.Before(time.Date(0o002, 1, 1, 0, 0, 0, 0, time.UTC)) {
 		return e.EncodeElement("", start)
 	}
-	return e.EncodeElement(t.Val.Format("2006-01-02T15:04:05"), start)
+	return e.EncodeElement(t.Val.Format(time.RFC3339), start)
 }
 
 func (t *SqlTimeStamp) UnmarshalXML(d *xml.Decoder, start xml.StartElement) error {
@@ -510,7 +511,7 @@ func (t *SqlTimeStamp) UnmarshalXML(d *xml.Decoder, start xml.StartElement) erro
 		return err
 	}
 	t.Val = tm
-	t.Valid = !tm.IsZero() && tm.Format("2006-01-02T15:04:05") != "0001-01-01T00:00:00"
+	t.Valid = !tm.IsZero() && tm.Format(time.RFC3339) != "0001-01-01T00:00:00Z"
 	return nil
 }
 

pkg/sqltypes/sql_types_test.go

@@ -178,7 +178,7 @@ func TestSqlTimeStamp_JSON(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Marshal failed: %v", err)
 	}
-	expected := `"2024-01-15T10:30:45"`
+	expected := `"2024-01-15T10:30:45Z"`
 	if string(data) != expected {
 		t.Errorf("expected %s, got %s", expected, string(data))
 	}
@@ -955,4 +955,3 @@ func TestSqlByteArray_Base64_RoundTrip(t *testing.T) {
 		t.Errorf("Round-trip failed: expected %v, got %v", original, b3.Val)
 	}
 }
-

pkg/sqltypes/sql_types_yaml_xml_test.go

@@ -195,7 +195,7 @@ func TestSqlTimeStamp_YAML(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Marshal failed: %v", err)
 	}
-	if string(data) != "2024-06-15T09:30:00\n" {
+	if string(data) != "\"2024-06-15T09:30:00Z\"\n" {
 		t.Errorf("unexpected YAML: %q", string(data))
 	}
 	var ts2 SqlTimeStamp

pkg/writers/bun/writer.go

@@ -195,7 +195,7 @@ func (w *Writer) writeMultiFile(db *models.Database) error {
 	}
 
 	// Create output directory if it doesn't exist
-	if err := os.MkdirAll(w.options.OutputPath, 0755); err != nil {
+	if err := os.MkdirAll(w.options.OutputPath, 0o755); err != nil {
 		return fmt.Errorf("failed to create output directory: %w", err)
 	}
 
@@ -267,7 +267,7 @@ func (w *Writer) writeMultiFile(db *models.Database) error {
 			filepath := filepath.Join(w.options.OutputPath, filename)
 
 			// Write file
-			if err := os.WriteFile(filepath, []byte(formatted), 0644); err != nil {
+			if err := os.WriteFile(filepath, []byte(formatted), 0o644); err != nil {
 				return fmt.Errorf("failed to write file %s: %w", filename, err)
 			}
 
@@ -471,7 +471,7 @@ func (w *Writer) formatCode(code string) (string, error) {
 // writeOutput writes the content to file or stdout
 func (w *Writer) writeOutput(content string) error {
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	// Print to stdout

pkg/writers/dbml/README.md

@@ -137,6 +137,41 @@ indexes {
 }
 ```
 
+### Dialect directives
+
+Dialect directives stored on a model object's `Metadata` (namespace `postgres`,
+`sqlite`, …) are re-emitted verbatim, one line per directive, at the location
+they belong to:
+
+```dbml
+@postgres: search_path myapp
+
+Table myapp.events {
+  id bigint [pk]
+  created_at timestamp [not null]
+  @postgres(id): identity always
+  @postgres: partition by RANGE (created_at)
+  @sqlite: without rowid
+
+  indexes {
+    (created_at) [name: 'idx_events_created']
+    @postgres: with (fillfactor=90)
+  }
+}
+```
+
+| Emitted at | From |
+|------------|------|
+| Before the first table | `Database.Metadata` |
+| After a column line, as `@ns(col): …` | `Column.Metadata` |
+| After an index line, inside `indexes { }` | `Index.Metadata` |
+| After the `indexes` block, before `Note:` | `Table.Metadata` |
+
+Output is deterministic (ordered by namespace, then source line, then args), so a
+`DBML → model → DBML` round-trip is idempotent. See
+[`docs/DBML_DIRECTIVES.md`](../../../docs/DBML_DIRECTIVES.md) for the grammar and
+the list of directives the PostgreSQL and SQLite writers translate to SQL.
+
 ## Type Mapping
 
 | SQL Type | DBML Type |

pkg/writers/dbml/directives.go

@@ -0,0 +1,23 @@
+package dbml
+
+import (
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+)
+
+// directiveLines renders every dialect directive stored in meta back to its DBML
+// source form, one line per directive, each prefixed with indent. When target is
+// non-empty it is emitted as the "(column)" target, e.g.
+// "  @postgres(id): identity always". Order is deterministic (see
+// models.GetDirectives).
+func directiveLines(meta map[string]any, indent, target string) []string {
+	directives := models.GetDirectives(meta)
+	if len(directives) == 0 {
+		return nil
+	}
+
+	lines := make([]string, 0, len(directives))
+	for _, d := range directives {
+		lines = append(lines, indent+models.FormatDirectiveLine(d, target))
+	}
+	return lines
+}

pkg/writers/dbml/directives_test.go

@@ -0,0 +1,97 @@
+package dbml
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	dbmlreader "git.warky.dev/wdevs/relspecgo/pkg/readers/dbml"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/readers"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers"
+)
+
+const directiveSrc = `@postgres: search_path myapp
+
+Table myapp.events {
+  id bigint [pk]
+  created_at timestamp [not null]
+  @postgres(id): identity always
+  @postgres: partition by RANGE (created_at)
+  @postgres: tablespace fast_data
+  @sqlite: without rowid
+
+  indexes {
+    (created_at) [name: 'idx_events_created']
+    @postgres: with (fillfactor=90)
+  }
+}
+`
+
+func writeDBML(t *testing.T, db *models.Database) string {
+	t.Helper()
+	out := filepath.Join(t.TempDir(), "out.dbml")
+	require.NoError(t, NewWriter(&writers.WriterOptions{OutputPath: out}).WriteDatabase(db))
+	b, err := os.ReadFile(out)
+	require.NoError(t, err)
+	return string(b)
+}
+
+func readDBML(t *testing.T, src string) *models.Database {
+	t.Helper()
+	f := filepath.Join(t.TempDir(), "in.dbml")
+	require.NoError(t, os.WriteFile(f, []byte(src), 0o644))
+	db, err := dbmlreader.NewReader(&readers.ReaderOptions{FilePath: f}).ReadDatabase()
+	require.NoError(t, err)
+	return db
+}
+
+func collectDirectives(db *models.Database) map[string][]string {
+	got := map[string][]string{}
+	add := func(loc string, meta map[string]any) {
+		for _, d := range models.GetDirectives(meta) {
+			got[loc] = append(got[loc], models.FormatDirectiveLine(d, ""))
+		}
+	}
+	add("database", db.Metadata)
+	for _, s := range db.Schemas {
+		for _, tbl := range s.Tables {
+			add("table:"+tbl.Name, tbl.Metadata)
+			for _, c := range tbl.Columns {
+				add("column:"+c.Name, c.Metadata)
+			}
+			for _, i := range tbl.Indexes {
+				add("index:"+i.Name, i.Metadata)
+			}
+		}
+	}
+	return got
+}
+
+func TestDirectives_RoundTrip(t *testing.T) {
+	db1 := readDBML(t, directiveSrc)
+	out1 := writeDBML(t, db1)
+	db2 := readDBML(t, out1)
+	out2 := writeDBML(t, db2)
+
+	assert.Equal(t, out1, out2, "DBML directive output should be idempotent")
+	assert.Equal(t, collectDirectives(db1), collectDirectives(db2), "directives preserved through round-trip")
+
+	// Spot-check each location survived.
+	d := collectDirectives(db2)
+	assert.Contains(t, d["database"], "@postgres: search_path myapp")
+	assert.Contains(t, d["table:events"], "@postgres: partition by RANGE (created_at)")
+	assert.Contains(t, d["table:events"], "@sqlite: without rowid")
+	assert.Contains(t, d["column:id"], "@postgres: identity always")
+	assert.Contains(t, d["index:idx_events_created"], "@postgres: with (fillfactor=90)")
+}
+
+func TestDirectives_WriterEmitsColumnTarget(t *testing.T) {
+	db := readDBML(t, directiveSrc)
+	out := writeDBML(t, db)
+	assert.Contains(t, out, "@postgres(id): identity always")
+}

pkg/writers/dbml/writer.go

@@ -27,7 +27,7 @@ func (w *Writer) WriteDatabase(db *models.Database) error {
 	content := w.databaseToDBML(db)
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	fmt.Print(content)
@@ -39,7 +39,7 @@ func (w *Writer) WriteSchema(schema *models.Schema) error {
 	content := w.schemaToDBML(schema)
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	fmt.Print(content)
@@ -51,7 +51,7 @@ func (w *Writer) WriteTable(table *models.Table) error {
 	content := w.tableToDBML(table)
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	fmt.Print(content)
@@ -72,6 +72,14 @@ func (w *Writer) databaseToDBML(d *models.Database) string {
 		sb.WriteString("\n")
 	}
 
+	if dirLines := directiveLines(d.Metadata, "", ""); len(dirLines) > 0 {
+		for _, line := range dirLines {
+			sb.WriteString(line)
+			sb.WriteString("\n")
+		}
+		sb.WriteString("\n")
+	}
+
 	for _, schema := range d.Schemas {
 		sb.WriteString(w.schemaToDBML(schema))
 	}
@@ -146,6 +154,11 @@ func (w *Writer) tableToDBML(t *models.Table) string {
 			fmt.Fprintf(&sb, " // %s", column.Comment)
 		}
 		sb.WriteString("\n")
+
+		for _, line := range directiveLines(column.Metadata, "  ", column.Name) {
+			sb.WriteString(line)
+			sb.WriteString("\n")
+		}
 	}
 
 	if len(t.Indexes) > 0 {
@@ -167,10 +180,20 @@ func (w *Writer) tableToDBML(t *models.Table) string {
 				fmt.Fprintf(&sb, " [%s]", strings.Join(indexAttrs, ", "))
 			}
 			sb.WriteString("\n")
+
+			for _, line := range directiveLines(index.Metadata, "    ", "") {
+				sb.WriteString(line)
+				sb.WriteString("\n")
+			}
 		}
 		sb.WriteString("  }\n")
 	}
 
+	for _, line := range directiveLines(t.Metadata, "  ", "") {
+		sb.WriteString(line)
+		sb.WriteString("\n")
+	}
+
 	note := strings.TrimSpace(t.Description + " " + t.Comment)
 	if note != "" {
 		fmt.Fprintf(&sb, "\n  Note: '%s'\n", note)

pkg/writers/dbml/writer_test.go

@@ -5,9 +5,10 @@ import (
 	"path/filepath"
 	"testing"
 
+	"github.com/stretchr/testify/assert"
+
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
 	"git.warky.dev/wdevs/relspecgo/pkg/writers"
-	"github.com/stretchr/testify/assert"
 )
 
 func TestWriter_WriteTable(t *testing.T) {

pkg/writers/dctx/roundtrip_test.go

@@ -5,11 +5,12 @@ import (
 	"path/filepath"
 	"testing"
 
+	"github.com/stretchr/testify/assert"
+
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
 	"git.warky.dev/wdevs/relspecgo/pkg/readers"
 	dctxreader "git.warky.dev/wdevs/relspecgo/pkg/readers/dctx"
 	"git.warky.dev/wdevs/relspecgo/pkg/writers"
-	"github.com/stretchr/testify/assert"
 )
 
 func TestRoundTrip_WriteAndRead(t *testing.T) {

pkg/writers/dctx/writer_test.go

@@ -5,9 +5,10 @@ import (
 	"os"
 	"testing"
 
+	"github.com/stretchr/testify/assert"
+
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
 	"git.warky.dev/wdevs/relspecgo/pkg/writers"
-	"github.com/stretchr/testify/assert"
 )
 
 func TestWriter_WriteSchema(t *testing.T) {

pkg/writers/drawdb/writer.go

@@ -48,7 +48,7 @@ func (w *Writer) writeJSON(data interface{}) error {
 	}
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, jsonData, 0644)
+		return os.WriteFile(w.options.OutputPath, jsonData, 0o644)
 	}
 
 	// If no output path, print to stdout

pkg/writers/drizzle/writer.go

@@ -115,7 +115,7 @@ func (w *Writer) writeMultiFile(db *models.Database) error {
 	}
 
 	// Create output directory if it doesn't exist
-	if err := os.MkdirAll(w.options.OutputPath, 0755); err != nil {
+	if err := os.MkdirAll(w.options.OutputPath, 0o755); err != nil {
 		return fmt.Errorf("failed to create output directory: %w", err)
 	}
 
@@ -163,7 +163,7 @@ func (w *Writer) writeEnumsFile(schema *models.Schema) error {
 
 	// Write to enums.ts file
 	filename := filepath.Join(w.options.OutputPath, "enums.ts")
-	return os.WriteFile(filename, []byte(code), 0644)
+	return os.WriteFile(filename, []byte(code), 0o644)
 }
 
 // writeTableFile writes a single table to its own file
@@ -200,7 +200,7 @@ func (w *Writer) writeTableFile(table *models.Table, schema *models.Schema, db *
 	// Sanitize table name to remove quotes, comments, and invalid characters
 	safeTableName := writers.SanitizeFilename(table.Name)
 	filename := filepath.Join(w.options.OutputPath, safeTableName+".ts")
-	return os.WriteFile(filename, []byte(code), 0644)
+	return os.WriteFile(filename, []byte(code), 0o644)
 }
 
 // buildTableData builds TableData from a models.Table
@@ -533,7 +533,7 @@ func (w *Writer) getForeignKeyForColumn(columnName string, table *models.Table)
 // writeOutput writes the content to file or stdout
 func (w *Writer) writeOutput(content string) error {
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	// Print to stdout

pkg/writers/gorm/writer.go

@@ -152,7 +152,7 @@ func (w *Writer) writeMultiFile(db *models.Database) error {
 	}
 
 	// Create output directory if it doesn't exist
-	if err := os.MkdirAll(w.options.OutputPath, 0755); err != nil {
+	if err := os.MkdirAll(w.options.OutputPath, 0o755); err != nil {
 		return fmt.Errorf("failed to create output directory: %w", err)
 	}
 
@@ -218,7 +218,7 @@ func (w *Writer) writeMultiFile(db *models.Database) error {
 			filepath := filepath.Join(w.options.OutputPath, filename)
 
 			// Write file
-			if err := os.WriteFile(filepath, []byte(formatted), 0644); err != nil {
+			if err := os.WriteFile(filepath, []byte(formatted), 0o644); err != nil {
 				return fmt.Errorf("failed to write file %s: %w", filename, err)
 			}
 
@@ -422,7 +422,7 @@ func (w *Writer) formatCode(code string) (string, error) {
 // writeOutput writes the content to file or stdout
 func (w *Writer) writeOutput(content string) error {
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	// Print to stdout

pkg/writers/graphql/relationships.go

@@ -76,7 +76,7 @@ func (w *Writer) generateRelationFields(table *models.Table, db *models.Database
 	return fields
 }
 
-func (w *Writer) getManyToManyField(table *models.Table, joinTable *models.Table, db *models.Database) string {
+func (w *Writer) getManyToManyField(table, joinTable *models.Table, db *models.Database) string {
 	// Find the two FK constraints in the join table
 	var fk1, fk2 *models.Constraint
 	for _, constraint := range joinTable.Constraints {

pkg/writers/graphql/writer.go

@@ -24,7 +24,7 @@ func (w *Writer) WriteDatabase(db *models.Database) error {
 	content := w.databaseToGraphQL(db)
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	fmt.Print(content)

pkg/writers/json/writer.go

@@ -55,7 +55,7 @@ func (w *Writer) WriteTable(table *models.Table) error {
 // writeOutput writes the content to file or stdout
 func (w *Writer) writeOutput(data []byte) error {
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, data, 0644)
+		return os.WriteFile(w.options.OutputPath, data, 0o644)
 	}
 
 	// Print to stdout

pkg/writers/mssql/writer_test.go

@@ -4,9 +4,10 @@ import (
 	"bytes"
 	"testing"
 
+	"github.com/stretchr/testify/assert"
+
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
 	"git.warky.dev/wdevs/relspecgo/pkg/writers"
-	"github.com/stretchr/testify/assert"
 )
 
 // TestGenerateColumnDefinition tests column definition generation

pkg/writers/pgsql/README.md

@@ -171,6 +171,27 @@ When `include_audit` is enabled, adds:
 - Function-based indexes
 - Concurrent index creation (`CREATE INDEX CONCURRENTLY`) via `Index.Concurrent`
 - Check constraints with expressions
+- Extension types and indexes: PostGIS, pgvector, citext, hstore, ltree (see below)
+- DBML dialect directives (`@postgres:` — see below)
+
+### DBML dialect directives
+
+`@postgres:` directives carried on a model object's `Metadata` (typically from a
+DBML source file) are translated to SQL:
+
+| Directive | Location | Emitted |
+|-----------|----------|---------|
+| `@postgres: partition by …` | table | `PARTITION BY …` on `CREATE TABLE` |
+| `@postgres: inherits …` | table | `INHERITS (…)` |
+| `@postgres: with (…)` | table, index | `WITH (…)` (on an index, overrides the comment-derived `WITH`) |
+| `@postgres: tablespace …` | table, index | `TABLESPACE …` |
+| `@postgres(col): storage …` | column | `STORAGE …` |
+| `@postgres(col): compression …` | column | `COMPRESSION …` |
+| `@postgres(col): identity always` / `identity by default` | column | `GENERATED ALWAYS/BY DEFAULT AS IDENTITY` |
+
+Directives for other dialects (`@sqlite:` …) are ignored. With
+`WriterOptions.StrictDirectives` (CLI `--strict-directives`) an untranslatable
+`@postgres:` key is an error. Full reference: [`docs/DBML_DIRECTIVES.md`](../../../docs/DBML_DIRECTIVES.md).
 
 ## Data Types
 
@@ -186,6 +207,108 @@ Supports all PostgreSQL data types:
 - Network: INET, CIDR, MACADDR
 - Special: ARRAY, HSTORE
 
+## Extension Types (PostGIS, pgvector)
+
+Extension column types are preserved verbatim, including their type modifier:
+
+| Type | Example column type | Extension |
+|------|---------------------|-----------|
+| PostGIS | `geometry(Point,4326)`, `geography(Point)`, `box2d`, `raster` | `postgis`, `postgis_raster`, `postgis_topology` |
+| pgvector | `vector(1536)`, `halfvec(768)`, `sparsevec(1000)` | `vector` |
+| Other | `citext`, `hstore`, `ltree` | `citext`, `hstore`, `ltree` |
+
+`CREATE EXTENSION IF NOT EXISTS <ext>;` is emitted automatically for every extension the
+schema needs. See [Extensions](#extensions).
+
+### Extension Indexes
+
+`Index.Type` selects the access method: `gist`, `spgist`, `brin` (PostGIS), `hnsw`, `ivfflat`
+(pgvector), `vchordrq`, `vchordg` (VectorChord), `bm25` (pg_search).
+
+Operator class and access-method parameters ride in `Index.Comment`:
+
+```
+opclass=vector_l2_ops; with (lists=100)
+```
+
+- `opclass=<name>` — used only when compatible with the column type; otherwise ignored.
+  Bare operator class names in the comment (e.g. `gin_trgm_ops`) are also recognized.
+- `with (k=v, …)` — rendered as `WITH (k = v, …)`. Only well-formed `key = value` pairs are
+  kept, so comment prose never reaches the DDL. Values may be bare (`lists=100`), quoted
+  (`key_field='id'`), or dollar-quoted (`options=$$[build.internal]$$`).
+
+Defaults when no operator class is requested:
+
+| Access method | Column type | Emitted operator class |
+|---------------|-------------|------------------------|
+| `hnsw`, `ivfflat`, `vchordrq`, `vchordg` | `vector` / `halfvec` / `sparsevec` / `bit` | `vector_cosine_ops` / `halfvec_cosine_ops` / `sparsevec_cosine_ops` / `bit_hamming_ops` |
+| `gist`, `spgist`, `brin` | `geometry`, `geography` | none (PostGIS default operator class) |
+| `gin` | text / `jsonb` / array | `gin_trgm_ops` / `jsonb_ops` / `array_ops` |
+
+pgvector defines no default operator class, so a vector index always names one.
+
+```sql
+CREATE INDEX IF NOT EXISTS idx_documents_embedding
+  ON public.documents USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100);
+CREATE INDEX IF NOT EXISTS idx_documents_location
+  ON public.documents USING gist (location);
+```
+
+Migrations only recreate an index when both sides specify a hint and they differ, so a model
+without hints does not churn against a live database.
+
+## Extensions
+
+`CREATE EXTENSION IF NOT EXISTS <ext>;` is emitted per schema, deduplicated and ordered so
+dependencies come first (`postgis` before `postgis_topology`/`postgis_raster`/`pgrouting`,
+`vector` before `vchord`). Names needing quoting are quoted: `CREATE EXTENSION IF NOT EXISTS "uuid-ossp";`
+
+### Detection
+
+| Source | Example | Extension |
+|--------|---------|-----------|
+| Column type | `vector(1536)`, `geometry(Point,4326)`, `citext`, `ltree` | `vector`, `postgis`, `citext`, `ltree` |
+| Index access method | `hnsw`, `ivfflat` / `vchordrq`, `vchordg` / `bm25` | `vector` / `vchord` / `pg_search` |
+| Operator class | `gin_trgm_ops`, `gist_ltree_ops` | `pg_trgm`, `ltree` |
+| GIN/GiST on a scalar type | `USING gin (views)` | `btree_gin` / `btree_gist` |
+| Function in a default, CHECK, index `WHERE`, or view body | `uuid_generate_v4()`, `crypt()`, `ST_Area()`, `unaccent()`, `json_matches_schema()` | `uuid-ossp`, `pgcrypto`, `postgis`, `unaccent`, `pg_jsonschema` |
+
+`gen_random_uuid()` is built in since PostgreSQL 13 and does not pull in `pgcrypto`.
+
+### Declaring extensions explicitly
+
+Extensions that leave no trace in the schema go in `schema.Metadata["extensions"]`, as a list
+or a comma-separated string. Dependencies are pulled in automatically; unknown names are kept
+as given. The PostgreSQL reader populates this from `pg_extension` for the schemas it reads.
+
+```yaml
+metadata:
+  extensions: [pg_cron, timescaledb, pg_stat_statements]
+```
+
+### Recognized extensions
+
+| Category | Extensions |
+|----------|------------|
+| ai/search | `vector`, `vchord` |
+| document | `hstore`, `ltree` |
+| federation | `postgres_fdw` |
+| geospatial | `postgis`, `postgis_raster`, `postgis_topology`, `pgrouting` |
+| indexing | `btree_gin`, `btree_gist` |
+| integration | `http` |
+| integrity | `amcheck` |
+| jobs / scheduling | `pg_background`, `pg_cron` |
+| maintenance | `pg_repack`, `pgstattuple` |
+| observability | `pg_qualstats`, `pg_stat_statements` |
+| partitioning | `pg_partman` |
+| procedural | `plpython3u` |
+| search | `pg_search`, `pg_textsearch` |
+| security | `pgcrypto` |
+| text | `citext`, `fuzzystrmatch`, `pg_trgm`, `unaccent` |
+| time-series | `timescaledb` |
+| utility | `uuid-ossp` |
+| validation | `pg_jsonschema` |
+
 ## Notes
 
 - Generated SQL is formatted and readable

pkg/writers/pgsql/directives.go

@@ -0,0 +1,184 @@
+package pgsql
+
+import (
+	"fmt"
+	"strings"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+)
+
+// directiveNamespace is the dialect namespace this writer consumes. Directives
+// for other namespaces (e.g. "sqlite") are ignored and never emitted as SQL.
+const directiveNamespace = "postgres"
+
+// pgHandledDirectives maps a directive location to the set of postgres keys this
+// writer knows how to translate. In strict mode an unknown key for this
+// namespace at a supported location is a hard error.
+var pgHandledDirectives = map[string]map[string]bool{
+	models.DirectiveLocationTable:  {"partition": true, "inherits": true, "with": true, "tablespace": true},
+	models.DirectiveLocationColumn: {"storage": true, "compression": true, "identity": true},
+	models.DirectiveLocationIndex:  {"with": true, "tablespace": true},
+}
+
+// checkDirectives validates postgres directives across a schema when strict mode
+// is enabled. It returns an error for any postgres directive whose key this
+// writer cannot translate. With strict mode off it is a no-op.
+func (w *Writer) checkDirectives(schema *models.Schema) error {
+	if w.options == nil || !w.options.StrictDirectives {
+		return nil
+	}
+	for _, table := range schema.Tables {
+		if err := checkObjectDirectives(table.Metadata, models.DirectiveLocationTable, table.Name); err != nil {
+			return err
+		}
+		for _, col := range table.Columns {
+			if err := checkObjectDirectives(col.Metadata, models.DirectiveLocationColumn, table.Name+"."+col.Name); err != nil {
+				return err
+			}
+		}
+		for _, idx := range table.Indexes {
+			if err := checkObjectDirectives(idx.Metadata, models.DirectiveLocationIndex, idx.Name); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+func checkObjectDirectives(meta map[string]any, location, owner string) error {
+	for _, d := range models.DirectivesForNamespace(meta, directiveNamespace) {
+		if !pgHandledDirectives[location][d.Key] {
+			return fmt.Errorf("pgsql: %s: unsupported @postgres directive %q at %s level (strict mode)", owner, d.Key, location)
+		}
+	}
+	return nil
+}
+
+// upperLeadingClause upcases a known leading keyword phrase in a directive
+// argument so the emitted SQL reads conventionally. Identifiers that follow are
+// left untouched.
+func upperLeadingClause(args, lowerPrefix, upperPrefix string) string {
+	args = strings.TrimSpace(args)
+	if strings.HasPrefix(strings.ToLower(args), lowerPrefix) {
+		return upperPrefix + args[len(lowerPrefix):]
+	}
+	return args
+}
+
+// pgTableDirectiveSuffix returns the clause appended after the closing ")" of a
+// CREATE TABLE statement, e.g. " PARTITION BY RANGE (created_at) TABLESPACE fast".
+func pgTableDirectiveSuffix(table *models.Table) string {
+	directives := models.DirectivesForNamespace(table.Metadata, directiveNamespace)
+	if len(directives) == 0 {
+		return ""
+	}
+
+	byKey := firstByKey(directives)
+
+	var parts []string
+	if d, ok := byKey["partition"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "partition by", "PARTITION BY"))
+	}
+	if d, ok := byKey["inherits"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "inherits", "INHERITS"))
+	}
+	if d, ok := byKey["with"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "with", "WITH"))
+	}
+	if d, ok := byKey["tablespace"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "tablespace", "TABLESPACE"))
+	}
+	if len(parts) == 0 {
+		return ""
+	}
+	return " " + strings.Join(parts, " ")
+}
+
+// pgColumnDirectiveSuffix returns the clause appended to a column definition,
+// e.g. " STORAGE PLAIN" or " GENERATED ALWAYS AS IDENTITY".
+func pgColumnDirectiveSuffix(col *models.Column) string {
+	directives := models.DirectivesForNamespace(col.Metadata, directiveNamespace)
+	if len(directives) == 0 {
+		return ""
+	}
+	byKey := firstByKey(directives)
+
+	var parts []string
+	if d, ok := byKey["storage"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "storage", "STORAGE"))
+	}
+	if d, ok := byKey["compression"]; ok {
+		parts = append(parts, upperLeadingClause(d.Args, "compression", "COMPRESSION"))
+	}
+	if d, ok := byKey["identity"]; ok {
+		parts = append(parts, identityClause(d.Args))
+	}
+	if len(parts) == 0 {
+		return ""
+	}
+	return " " + strings.Join(parts, " ")
+}
+
+// identityClause maps the two documented identity forms to standard SQL,
+// falling back to a verbatim (upcased-keyword) rendering.
+func identityClause(args string) string {
+	switch strings.ToLower(strings.Join(strings.Fields(args), " ")) {
+	case "identity always":
+		return "GENERATED ALWAYS AS IDENTITY"
+	case "identity default", "identity by default":
+		return "GENERATED BY DEFAULT AS IDENTITY"
+	default:
+		return upperLeadingClause(args, "identity", "IDENTITY")
+	}
+}
+
+// pgIndexDirectiveWith returns the parenthesised storage-parameter list from an
+// @postgres: with (...) index directive, e.g. "fillfactor=90", or "".
+func pgIndexDirectiveWith(index *models.Index) string {
+	for _, d := range models.DirectivesForNamespace(index.Metadata, directiveNamespace) {
+		if d.Key != "with" {
+			continue
+		}
+		inner := d.Args
+		if i := strings.Index(inner, "("); i >= 0 {
+			if j := strings.LastIndex(inner, ")"); j > i {
+				return strings.TrimSpace(inner[i+1 : j])
+			}
+		}
+		return strings.TrimSpace(strings.TrimPrefix(strings.ToLower(inner), "with"))
+	}
+	return ""
+}
+
+// pgIndexWithParams returns the storage-parameter list to use for an index,
+// preferring an @postgres: with (...) directive over the given fallback (e.g.
+// one derived from the index comment).
+func pgIndexWithParams(index *models.Index, fallback string) string {
+	if p := pgIndexDirectiveWith(index); p != "" {
+		return p
+	}
+	return fallback
+}
+
+// pgIndexDirectiveTablespace returns the tablespace name from an
+// @postgres: tablespace <name> index directive, or "".
+func pgIndexDirectiveTablespace(index *models.Index) string {
+	for _, d := range models.DirectivesForNamespace(index.Metadata, directiveNamespace) {
+		if d.Key == "tablespace" {
+			return strings.TrimSpace(strings.TrimPrefix(strings.ToLower(d.Args), "tablespace"))
+		}
+	}
+	return ""
+}
+
+// firstByKey indexes directives by key, keeping the first occurrence (the
+// documented postgres keys used here are all singletons).
+func firstByKey(directives []models.Directive) map[string]models.Directive {
+	byKey := make(map[string]models.Directive, len(directives))
+	for _, d := range directives {
+		if _, exists := byKey[d.Key]; !exists {
+			byKey[d.Key] = d
+		}
+	}
+	return byKey
+}

pkg/writers/pgsql/directives_test.go

@@ -0,0 +1,107 @@
+package pgsql
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers"
+)
+
+func directiveTestDB(t *testing.T) *models.Database {
+	t.Helper()
+	db := models.InitDatabase("testdb")
+	schema := models.InitSchema("public")
+	table := models.InitTable("events", "public")
+
+	id := models.InitColumn("id", "events", "public")
+	id.Type = "bigint"
+	id.IsPrimaryKey = true
+	id.NotNull = true
+	models.AddDirective(id.Metadata, models.Directive{Namespace: "postgres", Args: "identity always"})
+	models.AddDirective(id.Metadata, models.Directive{Namespace: "sqlite", Args: "collate NOCASE"})
+	table.Columns["id"] = id
+
+	created := models.InitColumn("created_at", "events", "public")
+	created.Type = "timestamp"
+	created.NotNull = true
+	table.Columns["created_at"] = created
+
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "postgres", Args: "partition by RANGE (created_at)"})
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "postgres", Args: "tablespace fast_data"})
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "sqlite", Args: "without rowid"})
+
+	idx := models.InitIndex("idx_events_created", "events", "public")
+	idx.Columns = []string{"created_at"}
+	models.AddDirective(idx.Metadata, models.Directive{Namespace: "postgres", Args: "with (fillfactor=90)"})
+	models.AddDirective(idx.Metadata, models.Directive{Namespace: "postgres", Args: "tablespace idx_space"})
+	table.Indexes["idx_events_created"] = idx
+
+	schema.Tables = append(schema.Tables, table)
+	db.Schemas = append(db.Schemas, schema)
+	return db
+}
+
+func TestPgDirectives_WriteDatabasePath(t *testing.T) {
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{})
+	w.writer = &buf
+	if err := w.WriteDatabase(directiveTestDB(t)); err != nil {
+		t.Fatalf("WriteDatabase: %v", err)
+	}
+	out := buf.String()
+
+	for _, want := range []string{
+		") PARTITION BY RANGE (created_at) TABLESPACE fast_data",
+		"GENERATED ALWAYS AS IDENTITY",
+		"WITH (fillfactor=90) TABLESPACE idx_space",
+	} {
+		if !strings.Contains(out, want) {
+			t.Errorf("missing %q in:\n%s", want, out)
+		}
+	}
+	// sqlite directives must never reach PG output.
+	if strings.Contains(out, "WITHOUT ROWID") || strings.Contains(strings.ToUpper(out), "COLLATE NOCASE") {
+		t.Errorf("sqlite directive leaked into PG output:\n%s", out)
+	}
+}
+
+func TestPgDirectives_WriteSchemaPath(t *testing.T) {
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{})
+	w.writer = &buf
+	if err := w.WriteSchema(directiveTestDB(t).Schemas[0]); err != nil {
+		t.Fatalf("WriteSchema: %v", err)
+	}
+	out := buf.String()
+	if !strings.Contains(out, ") PARTITION BY RANGE (created_at) TABLESPACE fast_data;") {
+		t.Errorf("table suffix missing from WriteSchema path:\n%s", out)
+	}
+	if !strings.Contains(out, "WITH (fillfactor=90) TABLESPACE idx_space") {
+		t.Errorf("index clauses missing from WriteSchema path:\n%s", out)
+	}
+}
+
+func TestPgDirectives_StrictUnknownKeyErrors(t *testing.T) {
+	db := directiveTestDB(t)
+	models.AddDirective(db.Schemas[0].Tables[0].Metadata, models.Directive{Namespace: "postgres", Args: "frobnicate x"})
+
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{StrictDirectives: true})
+	w.writer = &buf
+	err := w.WriteSchema(db.Schemas[0])
+	if err == nil || !strings.Contains(err.Error(), "frobnicate") {
+		t.Fatalf("want strict error for unknown postgres key, got %v", err)
+	}
+}
+
+func TestPgDirectives_StrictIgnoresOtherNamespaces(t *testing.T) {
+	// sqlite directives are present but must not trip PG strict mode.
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{StrictDirectives: true})
+	w.writer = &buf
+	if err := w.WriteSchema(directiveTestDB(t).Schemas[0]); err != nil {
+		t.Fatalf("strict mode should ignore non-postgres directives, got %v", err)
+	}
+}

pkg/writers/pgsql/extensions_test.go

@@ -0,0 +1,260 @@
+package pgsql
+
+import (
+	"reflect"
+	"strings"
+	"testing"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers"
+)
+
+// buildExtensionSchema returns a single-table schema the extension detection tests mutate.
+func buildExtensionSchema(t *testing.T) (*models.Schema, *models.Table) {
+	t.Helper()
+
+	schema := models.InitSchema("public")
+	table := models.InitTable("documents", "public")
+	schema.Tables = append(schema.Tables, table)
+	return schema, table
+}
+
+func addColumn(table *models.Table, name, sqlType string) *models.Column {
+	col := models.InitColumn(name, table.Name, table.Schema)
+	col.Type = sqlType
+	table.Columns[name] = col
+	return col
+}
+
+func TestRequiredExtensions_Detection(t *testing.T) {
+	tests := []struct {
+		name  string
+		build func(schema *models.Schema, table *models.Table)
+		want  []string
+	}{
+		{
+			name:  "no extensions",
+			build: func(_ *models.Schema, table *models.Table) { addColumn(table, "id", "integer") },
+			want:  nil,
+		},
+		{
+			name: "column type",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "embedding", "vector(1536)")
+				addColumn(table, "name", "citext")
+			},
+			want: []string{"citext", "vector"},
+		},
+		{
+			name: "column default function",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "id", "uuid").Default = "uuid_generate_v4()"
+			},
+			want: []string{"uuid-ossp"},
+		},
+		{
+			name: "check constraint expression",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "geom", "geometry")
+				table.Constraints["chk_geom"] = &models.Constraint{
+					Name:       "chk_geom",
+					Type:       models.CheckConstraint,
+					Expression: "ST_IsValid(geom)",
+				}
+			},
+			want: []string{"postgis"},
+		},
+		{
+			name: "partial index predicate",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "title", "text")
+				table.Indexes["idx_title"] = &models.Index{
+					Name:    "idx_title",
+					Type:    "btree",
+					Columns: []string{"title"},
+					Where:   "similarity(title, 'x') > 0.3",
+				}
+			},
+			want: []string{"pg_trgm"},
+		},
+		{
+			name: "view definition",
+			build: func(schema *models.Schema, table *models.Table) {
+				addColumn(table, "title", "text")
+				schema.Views = append(schema.Views, &models.View{
+					Name:       "v_documents",
+					Schema:     "public",
+					Definition: "SELECT unaccent(title) FROM documents",
+				})
+			},
+			want: []string{"unaccent"},
+		},
+		{
+			name: "index access method",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "body", "text")
+				table.Indexes["idx_body"] = &models.Index{
+					Name:    "idx_body",
+					Type:    "bm25",
+					Columns: []string{"body"},
+					Comment: "with (key_field='id')",
+				}
+			},
+			want: []string{"pg_search"},
+		},
+		{
+			name: "vchord depends on vector",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "embedding", "vector(3)")
+				table.Indexes["idx_embedding"] = &models.Index{
+					Name:    "idx_embedding",
+					Type:    "vchordrq",
+					Columns: []string{"embedding"},
+				}
+			},
+			want: []string{"vector", "vchord"},
+		},
+		{
+			name: "gin on scalar needs btree_gin",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "views", "integer")
+				table.Indexes["idx_views"] = &models.Index{
+					Name:    "idx_views",
+					Type:    "gin",
+					Columns: []string{"views"},
+				}
+			},
+			want: []string{"btree_gin"},
+		},
+		{
+			name: "gist on scalar needs btree_gist",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "views", "integer")
+				table.Indexes["idx_views"] = &models.Index{
+					Name:    "idx_views",
+					Type:    "gist",
+					Columns: []string{"views"},
+				}
+			},
+			want: []string{"btree_gist"},
+		},
+		{
+			name: "gist on geometry uses postgis operator classes",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "location", "geometry(Point,4326)")
+				table.Indexes["idx_location"] = &models.Index{
+					Name:    "idx_location",
+					Type:    "gist",
+					Columns: []string{"location"},
+				}
+			},
+			want: []string{"postgis"},
+		},
+		{
+			name: "gin on jsonb needs no companion",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "payload", "jsonb")
+				table.Indexes["idx_payload"] = &models.Index{
+					Name:    "idx_payload",
+					Type:    "gin",
+					Columns: []string{"payload"},
+				}
+			},
+			want: nil,
+		},
+		{
+			name: "gin on text uses pg_trgm",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "title", "text")
+				table.Indexes["idx_title"] = &models.Index{
+					Name:    "idx_title",
+					Type:    "gin",
+					Columns: []string{"title"},
+				}
+			},
+			want: []string{"pg_trgm"},
+		},
+		{
+			name: "gin on array needs no companion",
+			build: func(_ *models.Schema, table *models.Table) {
+				addColumn(table, "tags", "text[]")
+				table.Indexes["idx_tags"] = &models.Index{
+					Name:    "idx_tags",
+					Type:    "gin",
+					Columns: []string{"tags"},
+				}
+			},
+			want: nil,
+		},
+		{
+			name: "declared in metadata as string",
+			build: func(schema *models.Schema, _ *models.Table) {
+				schema.Metadata = map[string]any{"extensions": "pg_cron, timescaledb"}
+			},
+			want: []string{"pg_cron", "timescaledb"},
+		},
+		{
+			name: "declared in metadata as list",
+			build: func(schema *models.Schema, _ *models.Table) {
+				schema.Metadata = map[string]any{"extensions": []any{"postgis_topology", "pg_stat_statements"}}
+			},
+			// postgis is pulled in as a dependency of postgis_topology and emitted first.
+			want: []string{"pg_stat_statements", "postgis", "postgis_topology"},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			schema, table := buildExtensionSchema(t)
+			tt.build(schema, table)
+
+			if got := requiredExtensions(schema); !reflect.DeepEqual(got, tt.want) {
+				t.Errorf("requiredExtensions() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func TestRequiredExtensions_NilSchema(t *testing.T) {
+	if got := requiredExtensions(nil); got != nil {
+		t.Errorf("requiredExtensions(nil) = %v, want nil", got)
+	}
+}
+
+func TestWriteDatabase_QuotesExtensionNames(t *testing.T) {
+	db := models.InitDatabase("testdb")
+	schema, table := buildExtensionSchema(t)
+	addColumn(table, "id", "uuid").Default = "uuid_generate_v4()"
+	db.Schemas = append(db.Schemas, schema)
+
+	output := writeDatabaseOutput(t, db)
+	if !strings.Contains(output, `CREATE EXTENSION IF NOT EXISTS "uuid-ossp";`) {
+		t.Fatalf("expected quoted extension name, got:\n%s", output)
+	}
+}
+
+func TestGenerateSchemaStatements_ExtensionDependencyOrder(t *testing.T) {
+	schema, table := buildExtensionSchema(t)
+	addColumn(table, "embedding", "vector(3)")
+	table.Indexes["idx_embedding"] = &models.Index{
+		Name:    "idx_embedding",
+		Type:    "vchordrq",
+		Columns: []string{"embedding"},
+	}
+
+	writer := NewWriter(&writers.WriterOptions{})
+	statements, err := writer.GenerateSchemaStatements(schema)
+	if err != nil {
+		t.Fatalf("GenerateSchemaStatements failed: %v", err)
+	}
+
+	joined := strings.Join(statements, "\n")
+	vector := strings.Index(joined, "CREATE EXTENSION IF NOT EXISTS vector")
+	vchord := strings.Index(joined, "CREATE EXTENSION IF NOT EXISTS vchord")
+	if vector < 0 || vchord < 0 {
+		t.Fatalf("expected vector and vchord extensions, got:\n%s", joined)
+	}
+	if vector > vchord {
+		t.Fatalf("expected vector to be created before vchord, got:\n%s", joined)
+	}
+}

pkg/writers/pgsql/migration_writer.go

@@ -47,7 +47,7 @@ func NewMigrationWriter(options *writers.WriterOptions) (*MigrationWriter, error
 }
 
 // WriteMigration generates migration scripts using templates
-func (w *MigrationWriter) WriteMigration(model *models.Database, current *models.Database) error {
+func (w *MigrationWriter) WriteMigration(model, current *models.Database) error {
 	if model == nil {
 		return fmt.Errorf("model database is required")
 	}
@@ -161,17 +161,17 @@ func (w *MigrationWriter) WriteMigration(model *models.Database, current *models
 }
 
 // generateSchemaScripts generates migration scripts for a schema using templates
-func (w *MigrationWriter) generateSchemaScripts(model *models.Schema, current *models.Schema) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateSchemaScripts(model, current *models.Schema) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 
-	if schemaRequiresPGTrgm(model) {
+	for _, extension := range requiredExtensions(model) {
 		scripts = append(scripts, MigrationScript{
-			ObjectName: "extension.pg_trgm",
+			ObjectName: "extension." + extension,
 			ObjectType: "create extension",
 			Schema:     model.Name,
 			Priority:   80,
 			Sequence:   len(scripts),
-			Body:       "CREATE EXTENSION IF NOT EXISTS pg_trgm;",
+			Body:       fmt.Sprintf("CREATE EXTENSION IF NOT EXISTS %s;", pgsql.QuoteExtensionName(extension)),
 		})
 	}
 
@@ -220,7 +220,7 @@ func (w *MigrationWriter) generateSchemaScripts(model *models.Schema, current *m
 // generateDropScripts generates DROP scripts using templates.
 // Returns the scripts and a set of FK constraint keys (schema.table.name) that were
 // explicitly dropped because their referenced PK was being dropped, so they can be force-recreated.
-func (w *MigrationWriter) generateDropScripts(model *models.Schema, current *models.Schema) ([]MigrationScript, map[string]bool, error) {
+func (w *MigrationWriter) generateDropScripts(model, current *models.Schema) ([]MigrationScript, map[string]bool, error) {
 	scripts := make([]MigrationScript, 0)
 	droppedFKs := make(map[string]bool)
 
@@ -349,7 +349,7 @@ func (w *MigrationWriter) generateDropScripts(model *models.Schema, current *mod
 }
 
 // generateTableScripts generates CREATE/ALTER TABLE scripts using templates
-func (w *MigrationWriter) generateTableScripts(model *models.Schema, current *models.Schema) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateTableScripts(model, current *models.Schema) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 
 	// Build map of current tables
@@ -394,7 +394,7 @@ func (w *MigrationWriter) generateTableScripts(model *models.Schema, current *mo
 }
 
 // generateAlterTableScripts generates ALTER TABLE scripts using templates
-func (w *MigrationWriter) generateAlterTableScripts(schema *models.Schema, modelTable *models.Table, currentTable *models.Table) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateAlterTableScripts(schema *models.Schema, modelTable, currentTable *models.Table) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 
 	// Build map of current columns
@@ -514,7 +514,7 @@ func (w *MigrationWriter) generateAlterTableScripts(schema *models.Schema, model
 }
 
 // generateIndexScripts generates CREATE INDEX scripts using templates
-func (w *MigrationWriter) generateIndexScripts(model *models.Schema, current *models.Schema) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateIndexScripts(model, current *models.Schema) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 
 	// Build map of current tables
@@ -653,6 +653,7 @@ func (w *MigrationWriter) generateIndexScripts(model *models.Schema, current *mo
 					Columns:           strings.Join(columnExprs, ", "),
 					Unique:            modelIndex.Unique,
 					Concurrent:        modelIndex.Concurrent,
+					StorageParameters: indexStorageParameters(modelIndex.Comment),
 				})
 				if err != nil {
 					return nil, err
@@ -674,20 +675,31 @@ func (w *MigrationWriter) generateIndexScripts(model *models.Schema, current *mo
 	return scripts, nil
 }
 
+// buildIndexColumnExpressions renders the column list of an index, appending the operator
+// class each column needs for the access method (GIN opclasses, pgvector distance ops,
+// explicitly requested PostGIS opclasses). Columns that cannot be resolved on the table are
+// emitted verbatim.
 func buildIndexColumnExpressions(table *models.Table, index *models.Index, indexType string) []string {
+	return buildIndexColumnExpressionsFiltered(table, index, indexType, false)
+}
+
+// buildIndexColumnExpressionsFiltered is buildIndexColumnExpressions with the option to drop
+// columns that do not exist on the table instead of emitting them verbatim.
+func buildIndexColumnExpressionsFiltered(table *models.Table, index *models.Index, indexType string, skipUnresolved bool) []string {
 	columnExprs := make([]string, 0, len(index.Columns))
 	for _, colName := range index.Columns {
-		colExpr := colName
-		if table != nil {
-			if col, ok := resolveIndexColumn(table, colName); ok && col != nil {
-				colExpr = col.SQLName()
-				if strings.EqualFold(indexType, "gin") {
-					opClass := ginOperatorClassForColumn(col, index.Comment)
-					if opClass != "" {
-						colExpr = fmt.Sprintf("%s %s", col.SQLName(), opClass)
-					}
+		col, ok := resolveIndexColumn(table, colName)
+		if !ok || col == nil {
+			if skipUnresolved {
+				continue
 			}
+			columnExprs = append(columnExprs, colName)
+			continue
 		}
+
+		colExpr := col.SQLName()
+		if opClass := indexOperatorClassForColumn(col, indexType, index.Comment); opClass != "" {
+			colExpr = fmt.Sprintf("%s %s", colExpr, opClass)
 		}
 		columnExprs = append(columnExprs, colExpr)
 	}
@@ -697,7 +709,7 @@ func buildIndexColumnExpressions(table *models.Table, index *models.Index, index
 // generateForeignKeyScripts generates ADD CONSTRAINT FOREIGN KEY scripts using templates.
 // forceRecreate is a set of FK constraint keys (schema.table.name) that must be recreated
 // even if unchanged, because their referenced PK was dropped and recreated.
-func (w *MigrationWriter) generateForeignKeyScripts(model *models.Schema, current *models.Schema, forceRecreate map[string]bool) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateForeignKeyScripts(model, current *models.Schema, forceRecreate map[string]bool) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 
 	// Build map of current tables
@@ -775,7 +787,7 @@ func (w *MigrationWriter) generateForeignKeyScripts(model *models.Schema, curren
 }
 
 // generateCommentScripts generates COMMENT ON scripts using templates
-func (w *MigrationWriter) generateCommentScripts(model *models.Schema, current *models.Schema) ([]MigrationScript, error) {
+func (w *MigrationWriter) generateCommentScripts(model, current *models.Schema) ([]MigrationScript, error) {
 	scripts := make([]MigrationScript, 0)
 	_ = current // TODO: Compare with current schema to only add new/changed comments
 
@@ -1046,5 +1058,19 @@ func indexesEqual(idx1, idx2 *models.Index) bool {
 			return false
 		}
 	}
+	// Operator class and storage parameters ride along in the index comment. They only
+	// signal a difference when both sides specify one, so an index whose model side omits
+	// the hint is not recreated on every migration.
+	if !indexHintsEqual(extractOperatorClass(idx1.Comment), extractOperatorClass(idx2.Comment)) {
+		return false
+	}
+	return indexHintsEqual(indexStorageParameters(idx1.Comment), indexStorageParameters(idx2.Comment))
+}
+
+// indexHintsEqual compares two optional index hints, treating an unspecified hint as a match.
+func indexHintsEqual(hint1, hint2 string) bool {
+	if hint1 == "" || hint2 == "" {
 		return true
 	}
+	return strings.EqualFold(hint1, hint2)
+}

pkg/writers/pgsql/migration_writer_test.go

@@ -852,3 +852,93 @@ func TestWriteMigration_NilCurrentTreatsDatabaseAsEmpty(t *testing.T) {
 		t.Fatalf("expected CREATE TABLE in migration output, got:\n%s", output)
 	}
 }
+
+func TestWriteMigration_VectorAndPostGISIndexes(t *testing.T) {
+	current := models.InitDatabase("testdb")
+	current.Schemas = append(current.Schemas, models.InitSchema("public"))
+
+	model := models.InitDatabase("testdb")
+	modelSchema := models.InitSchema("public")
+
+	table := models.InitTable("documents", "public")
+
+	embedding := models.InitColumn("embedding", "documents", "public")
+	embedding.Type = "vector(1536)"
+	table.Columns["embedding"] = embedding
+
+	location := models.InitColumn("location", "documents", "public")
+	location.Type = "geometry(Point,4326)"
+	table.Columns["location"] = location
+
+	table.Indexes["idx_documents_embedding"] = &models.Index{
+		Name:    "idx_documents_embedding",
+		Type:    "ivfflat",
+		Columns: []string{"embedding"},
+		Comment: "opclass=vector_cosine_ops; with (lists=100)",
+	}
+	table.Indexes["idx_documents_location"] = &models.Index{
+		Name:    "idx_documents_location",
+		Type:    "gist",
+		Columns: []string{"location"},
+	}
+
+	modelSchema.Tables = append(modelSchema.Tables, table)
+	model.Schemas = append(model.Schemas, modelSchema)
+
+	var buf bytes.Buffer
+	writer, err := NewMigrationWriter(&writers.WriterOptions{})
+	if err != nil {
+		t.Fatalf("Failed to create writer: %v", err)
+	}
+	writer.writer = &buf
+
+	if err := writer.WriteMigration(model, current); err != nil {
+		t.Fatalf("WriteMigration failed: %v", err)
+	}
+
+	output := buf.String()
+	for _, want := range []string{
+		"CREATE EXTENSION IF NOT EXISTS postgis;",
+		"CREATE EXTENSION IF NOT EXISTS vector;",
+		"vector(1536)",
+		"geometry(Point,4326)",
+		"USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100)",
+		"USING gist (location)",
+	} {
+		if !strings.Contains(output, want) {
+			t.Fatalf("expected migration to contain %q, got:\n%s", want, output)
+		}
+	}
+}
+
+func TestIndexesEqual_OperatorClassAndStorageParameters(t *testing.T) {
+	newIndex := func(comment string) *models.Index {
+		return &models.Index{
+			Name:    "idx_documents_embedding",
+			Type:    "hnsw",
+			Columns: []string{"embedding"},
+			Comment: comment,
+		}
+	}
+
+	tests := []struct {
+		name      string
+		comment1  string
+		comment2  string
+		wantEqual bool
+	}{
+		{"identical hints", "opclass=vector_l2_ops", "opclass=vector_l2_ops", true},
+		{"different operator class", "opclass=vector_l2_ops", "opclass=vector_cosine_ops", false},
+		{"different storage parameters", "with (m=16)", "with (m=32)", false},
+		{"unspecified hint on one side", "", "opclass=vector_l2_ops; with (m=16)", true},
+		{"unrelated comments", "primary lookup index", "primary lookup index", true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := indexesEqual(newIndex(tt.comment1), newIndex(tt.comment2)); got != tt.wantEqual {
+				t.Errorf("indexesEqual() = %v, want %v", got, tt.wantEqual)
+			}
+		})
+	}
+}

pkg/writers/pgsql/templates.go

@@ -140,6 +140,9 @@ type CreateIndexData struct {
 	Columns    string
 	Unique     bool
 	Concurrent bool
+	// StorageParameters holds access-method parameters rendered as WITH (...),
+	// e.g. "lists = 100" for ivfflat or "m = 16, ef_construction = 64" for hnsw.
+	StorageParameters string
 }
 
 // CreateForeignKeyData contains data for create foreign key template

pkg/writers/pgsql/templates/create_index.tmpl

@@ -1,2 +1,2 @@
 CREATE {{if .Unique}}UNIQUE {{end}}INDEX {{if .Concurrent}}CONCURRENTLY {{end}}IF NOT EXISTS {{quote_ident .IndexName}}
-  ON {{qual_table .SchemaName .TableName}} USING {{.IndexType}} ({{.Columns}});
+  ON {{qual_table .SchemaName .TableName}} USING {{.IndexType}} ({{.Columns}}){{if .StorageParameters}} WITH ({{.StorageParameters}}){{end}};

pkg/writers/pgsql/writer.go

@@ -6,8 +6,10 @@ import (
 	"fmt"
 	"io"
 	"os"
+	"regexp"
 	"sort"
 	"strings"
+	"sync"
 	"time"
 
 	"git.warky.dev/wdevs/relspecgo/pkg/models"
@@ -141,14 +143,18 @@ func (w *Writer) GenerateDatabaseStatements(db *models.Database) ([]string, erro
 func (w *Writer) GenerateSchemaStatements(schema *models.Schema) ([]string, error) {
 	statements := []string{}
 
+	if err := w.checkDirectives(schema); err != nil {
+		return nil, err
+	}
+
 	// Phase 1: Create schema (skip entirely when flattening)
 	if schema.Name != "public" && !w.options.FlattenSchema {
 		statements = append(statements, fmt.Sprintf("-- Schema: %s", schema.Name))
 		statements = append(statements, fmt.Sprintf("CREATE SCHEMA IF NOT EXISTS %s", schema.SQLName()))
 	}
 
-	if schemaRequiresPGTrgm(schema) {
-		statements = append(statements, `CREATE EXTENSION IF NOT EXISTS pg_trgm`)
+	for _, extension := range requiredExtensions(schema) {
+		statements = append(statements, fmt.Sprintf("CREATE EXTENSION IF NOT EXISTS %s", pgsql.QuoteExtensionName(extension)))
 	}
 
 	// Phase 2: Create sequences
@@ -271,18 +277,17 @@ func (w *Writer) GenerateSchemaStatements(schema *models.Schema) ([]string, erro
 				indexType = "btree"
 			}
 
-			// Build column expressions with operator class support for GIN indexes
-			columnExprs := make([]string, 0, len(index.Columns))
-			for _, colName := range index.Columns {
-				colExpr := colName
-				if col, ok := resolveIndexColumn(table, colName); ok {
-					if strings.EqualFold(indexType, "gin") {
-						if opClass := ginOperatorClassForColumn(col, index.Comment); opClass != "" {
-							colExpr = fmt.Sprintf("%s %s", colName, opClass)
+			// Build column expressions with operator class support (GIN, pgvector, PostGIS)
+			columnExprs := buildIndexColumnExpressions(table, index, indexType)
+
+			withClause := ""
+			if params := pgIndexWithParams(index, indexStorageParameters(index.Comment)); params != "" {
+				withClause = fmt.Sprintf(" WITH (%s)", params)
 			}
-					}
-				}
-				columnExprs = append(columnExprs, colExpr)
+
+			tablespaceClause := ""
+			if ts := pgIndexDirectiveTablespace(index); ts != "" {
+				tablespaceClause = fmt.Sprintf(" TABLESPACE %s", ts)
 			}
 
 			whereClause := ""
@@ -290,8 +295,8 @@ func (w *Writer) GenerateSchemaStatements(schema *models.Schema) ([]string, erro
 				whereClause = fmt.Sprintf(" WHERE %s", index.Where)
 			}
 
-			stmt := fmt.Sprintf("CREATE %sINDEX IF NOT EXISTS %s ON %s USING %s (%s)%s",
-				uniqueStr, quoteIdentifier(index.Name), w.qualTable(schema.SQLName(), table.SQLName()), indexType, strings.Join(columnExprs, ", "), whereClause)
+			stmt := fmt.Sprintf("CREATE %sINDEX IF NOT EXISTS %s ON %s USING %s (%s)%s%s%s",
+				uniqueStr, quoteIdentifier(index.Name), w.qualTable(schema.SQLName(), table.SQLName()), indexType, strings.Join(columnExprs, ", "), withClause, tablespaceClause, whereClause)
 			statements = append(statements, stmt)
 		}
 	}
@@ -585,8 +590,9 @@ func (w *Writer) generateCreateTableStatement(schema *models.Schema, table *mode
 		columnDefs = append(columnDefs, "  "+def)
 	}
 
-	stmt := fmt.Sprintf("CREATE TABLE IF NOT EXISTS %s (\n%s\n)",
-		w.qualTable(schema.SQLName(), table.SQLName()), strings.Join(columnDefs, ",\n"))
+	stmt := fmt.Sprintf("CREATE TABLE IF NOT EXISTS %s (\n%s\n)%s",
+		w.qualTable(schema.SQLName(), table.SQLName()), strings.Join(columnDefs, ",\n"),
+		pgTableDirectiveSuffix(table))
 	statements = append(statements, stmt)
 
 	return statements, nil
@@ -615,7 +621,7 @@ func (w *Writer) generateColumnDefinition(col *models.Column) string {
 		}
 	}
 
-	return strings.Join(parts, " ")
+	return strings.Join(parts, " ") + pgColumnDirectiveSuffix(col)
 }
 
 func effectiveColumnSQLType(col *models.Column) string {
@@ -682,6 +688,10 @@ func (w *Writer) WriteSchema(schema *models.Schema) error {
 		w.writer = os.Stdout
 	}
 
+	if err := w.checkDirectives(schema); err != nil {
+		return err
+	}
+
 	// Phase 1: Create schema (priority 1)
 	if err := w.writeCreateSchema(schema); err != nil {
 		return err
@@ -819,11 +829,14 @@ func (w *Writer) writeCreateSchema(schema *models.Schema) error {
 }
 
 func (w *Writer) writeRequiredExtensions(schema *models.Schema) error {
-	if !schemaRequiresPGTrgm(schema) {
+	extensions := requiredExtensions(schema)
+	if len(extensions) == 0 {
 		return nil
 	}
 
-	fmt.Fprintln(w.writer, "CREATE EXTENSION IF NOT EXISTS pg_trgm;")
+	for _, extension := range extensions {
+		fmt.Fprintf(w.writer, "CREATE EXTENSION IF NOT EXISTS %s;\n", pgsql.QuoteExtensionName(extension))
+	}
 	fmt.Fprintln(w.writer)
 	return nil
 }
@@ -885,7 +898,7 @@ func (w *Writer) writeCreateTables(schema *models.Schema) error {
 		}
 
 		fmt.Fprintf(w.writer, "%s\n", strings.Join(columnDefs, ",\n"))
-		fmt.Fprintf(w.writer, ");\n\n")
+		fmt.Fprintf(w.writer, ")%s;\n\n", pgTableDirectiveSuffix(table))
 	}
 
 	return nil
@@ -1063,21 +1076,13 @@ func (w *Writer) writeIndexes(schema *models.Schema) error {
 				indexName = fmt.Sprintf("%s_%s_%s", indexType, table.SQLName(), strings.ToLower(columnSuffix))
 			}
 
-			// Build column list with operator class support for GIN indexes
-			columnExprs := make([]string, 0, len(index.Columns))
-			for _, colName := range index.Columns {
-				if col, ok := resolveIndexColumn(table, colName); ok {
-					colExpr := col.SQLName()
-					if strings.EqualFold(index.Type, "gin") {
-						opClass := ginOperatorClassForColumn(col, index.Comment)
-						if opClass != "" {
-							colExpr = fmt.Sprintf("%s %s", col.SQLName(), opClass)
-						}
-					}
-					columnExprs = append(columnExprs, colExpr)
-				}
+			indexType := index.Type
+			if indexType == "" {
+				indexType = "btree"
 			}
 
+			// Build column list with operator class support (GIN, pgvector, PostGIS)
+			columnExprs := buildIndexColumnExpressionsFiltered(table, index, indexType, true)
 			if len(columnExprs) == 0 {
 				continue
 			}
@@ -1087,9 +1092,14 @@ func (w *Writer) writeIndexes(schema *models.Schema) error {
 				unique = "UNIQUE "
 			}
 
-			indexType := index.Type
-			if indexType == "" {
-				indexType = "btree"
+			withClause := ""
+			if params := pgIndexWithParams(index, indexStorageParameters(index.Comment)); params != "" {
+				withClause = fmt.Sprintf(" WITH (%s)", params)
+			}
+
+			tablespaceClause := ""
+			if ts := pgIndexDirectiveTablespace(index); ts != "" {
+				tablespaceClause = fmt.Sprintf(" TABLESPACE %s", ts)
 			}
 
 			whereClause := ""
@@ -1104,8 +1114,8 @@ func (w *Writer) writeIndexes(schema *models.Schema) error {
 
 			fmt.Fprintf(w.writer, "CREATE %sINDEX %sIF NOT EXISTS %s\n",
 				unique, concurrently, indexName)
-			fmt.Fprintf(w.writer, "  ON %s USING %s (%s)%s;\n\n",
-				w.qualTable(schema.SQLName(), table.SQLName()), indexType, strings.Join(columnExprs, ", "), whereClause)
+			fmt.Fprintf(w.writer, "  ON %s USING %s (%s)%s%s%s;\n\n",
+				w.qualTable(schema.SQLName(), table.SQLName()), indexType, strings.Join(columnExprs, ", "), withClause, tablespaceClause, whereClause)
 		}
 	}
 
@@ -1483,7 +1493,69 @@ func isTextTypeWithoutLength(colType string) bool {
 	return strings.EqualFold(colType, "text")
 }
 
-func ginOperatorClassForColumn(col *models.Column, comment string) string {
+// vectorOperatorClasses maps pgvector operator classes to the column base type they
+// apply to. pgvector defines no default operator class, so an hnsw/ivfflat index must
+// always name one explicitly.
+var vectorOperatorClasses = map[string]string{
+	"vector_l2_ops":        "vector",
+	"vector_ip_ops":        "vector",
+	"vector_cosine_ops":    "vector",
+	"vector_l1_ops":        "vector",
+	"halfvec_l2_ops":       "halfvec",
+	"halfvec_ip_ops":       "halfvec",
+	"halfvec_cosine_ops":   "halfvec",
+	"halfvec_l1_ops":       "halfvec",
+	"sparsevec_l2_ops":     "sparsevec",
+	"sparsevec_ip_ops":     "sparsevec",
+	"sparsevec_cosine_ops": "sparsevec",
+	"sparsevec_l1_ops":     "sparsevec",
+	"bit_hamming_ops":      "bit",
+	"bit_jaccard_ops":      "bit",
+}
+
+// defaultVectorOperatorClasses is the operator class used for an hnsw/ivfflat index when
+// the index comment does not request one. Cosine distance is the common default for
+// embedding columns; override it with an "opclass" hint in the index comment.
+var defaultVectorOperatorClasses = map[string]string{
+	"vector":    "vector_cosine_ops",
+	"halfvec":   "halfvec_cosine_ops",
+	"sparsevec": "sparsevec_cosine_ops",
+	"bit":       "bit_hamming_ops",
+}
+
+// spatialOperatorClasses are the PostGIS operator classes recognized in index comments.
+// PostGIS installs default operator classes for gist/spgist/brin, so these are only
+// emitted when explicitly requested (e.g. the 3D/nD variants).
+var spatialOperatorClasses = map[string]bool{
+	"gist_geometry_ops_2d":            true,
+	"gist_geometry_ops_nd":            true,
+	"gist_geography_ops":              true,
+	"spgist_geometry_ops_2d":          true,
+	"spgist_geometry_ops_3d":          true,
+	"spgist_geometry_ops_nd":          true,
+	"brin_geometry_inclusion_ops_2d":  true,
+	"brin_geometry_inclusion_ops_3d":  true,
+	"brin_geometry_inclusion_ops_4d":  true,
+	"brin_geography_inclusion_ops_2d": true,
+	"btree_geometry_ops":              true,
+	"btree_geography_ops":             true,
+}
+
+// isVectorIndexMethod reports whether the access method indexes pgvector types, which
+// covers both pgvector itself (hnsw, ivfflat) and VectorChord (vchordrq, vchordg).
+func isVectorIndexMethod(method string) bool {
+	switch strings.ToLower(strings.TrimSpace(method)) {
+	case "hnsw", "ivfflat", "vchordrq", "vchordg":
+		return true
+	default:
+		return false
+	}
+}
+
+// indexOperatorClassForColumn returns the operator class to emit for a column in an index
+// of the given access method, honouring an explicit request from the index comment when it
+// is compatible with the column type.
+func indexOperatorClassForColumn(col *models.Column, indexType, comment string) string {
 	if col == nil {
 		return ""
 	}
@@ -1492,15 +1564,20 @@ func ginOperatorClassForColumn(col *models.Column, comment string) string {
 	baseType := pgsql.CanonicalizeBaseType(pgsql.ExtractBaseTypeLower(sqlType))
 	isArray := pgsql.IsArrayType(sqlType)
 	requested := extractOperatorClass(comment)
+	method := strings.ToLower(strings.TrimSpace(indexType))
+	if method == "" {
+		method = "btree"
+	}
 
-	if requested != "" && ginOperatorClassCompatible(baseType, isArray, requested) {
+	if requested != "" && operatorClassCompatible(method, baseType, isArray, requested) {
 		return requested
 	}
 
+	switch {
+	case method == "gin":
 		if isArray {
 			return "array_ops"
 		}
-
 		switch {
 		case isTextGinBaseType(baseType):
 			return "gin_trgm_ops"
@@ -1509,9 +1586,26 @@ func ginOperatorClassForColumn(col *models.Column, comment string) string {
 		default:
 			return requested
 		}
+	case isVectorIndexMethod(method):
+		if isArray {
+			return ""
+		}
+		return defaultVectorOperatorClasses[baseType]
+	default:
+		// gist/spgist/brin/btree have default operator classes (PostGIS included),
+		// so nothing is emitted unless the comment requested a compatible class.
+		return ""
+	}
+}
+
+func operatorClassCompatible(method, baseType string, isArray bool, opClass string) bool {
+	if vectorType, ok := vectorOperatorClasses[opClass]; ok {
+		return !isArray && baseType == vectorType && isVectorIndexMethod(method)
+	}
+	if spatialOperatorClasses[opClass] {
+		return !isArray && pgsql.IsSpatialType(baseType)
 	}
 
-func ginOperatorClassCompatible(baseType string, isArray bool, opClass string) bool {
 	switch opClass {
 	case "gin_trgm_ops", "gin_bigm_ops":
 		return !isArray && isTextGinBaseType(baseType)
@@ -1533,30 +1627,180 @@ func isTextGinBaseType(baseType string) bool {
 	}
 }
 
-func schemaRequiresPGTrgm(schema *models.Schema) bool {
+// requiredExtensions returns the PostgreSQL extensions a schema depends on, ordered so
+// that dependencies are created first (postgis before postgis_topology, vector before
+// vchord). Extensions are detected from column types, index access methods, resolved
+// operator classes, and function calls in defaults, check constraints, partial index
+// predicates and view definitions. Extensions that leave no trace in the model (pg_cron,
+// timescaledb, postgres_fdw, …) can be declared in schema.Metadata["extensions"].
+func requiredExtensions(schema *models.Schema) []string {
 	if schema == nil {
-		return false
+		return nil
 	}
+
+	required := make(map[string]bool)
+	add := func(names ...string) {
+		for _, name := range names {
+			if name != "" {
+				required[name] = true
+			}
+		}
+	}
+
+	add(declaredExtensions(schema)...)
+
+	for _, view := range schema.Views {
+		if view == nil {
+			continue
+		}
+		add(pgsql.ExtensionsForExpression(view.Definition)...)
+	}
+
 	for _, table := range schema.Tables {
 		if table == nil {
 			continue
 		}
-		for _, index := range table.Indexes {
-			if index == nil || !strings.EqualFold(index.Type, "gin") {
+
+		for _, col := range table.Columns {
+			if col == nil {
 				continue
 			}
+			add(pgsql.TypeExtension(effectiveColumnSQLType(col)))
+			if def, ok := col.Default.(string); ok {
+				add(pgsql.ExtensionsForExpression(def)...)
+			}
+		}
+
+		for _, constraint := range table.Constraints {
+			if constraint == nil {
+				continue
+			}
+			add(pgsql.ExtensionsForExpression(constraint.Expression)...)
+		}
+
+		for _, index := range table.Indexes {
+			if index == nil {
+				continue
+			}
+			add(pgsql.IndexMethodExtension(index.Type))
+			add(pgsql.ExtensionsForExpression(index.Where)...)
+
 			for _, colName := range index.Columns {
 				col, ok := resolveIndexColumn(table, colName)
 				if !ok || col == nil {
 					continue
 				}
-				if ginOperatorClassForColumn(col, index.Comment) == "gin_trgm_ops" {
+				opClass := indexOperatorClassForColumn(col, index.Type, index.Comment)
+				add(pgsql.OperatorClassExtension(opClass))
+				add(btreeCompanionExtension(index.Type, col, opClass))
+			}
+		}
+	}
+
+	extensions := make([]string, 0, len(required))
+	for ext := range required {
+		extensions = append(extensions, ext)
+	}
+
+	// Pull in dependencies, so a declared postgis_topology also creates postgis.
+	for i := 0; i < len(extensions); i++ {
+		for _, dependency := range pgsql.ExtensionDependencies(extensions[i]) {
+			if !required[dependency] {
+				required[dependency] = true
+				extensions = append(extensions, dependency)
+			}
+		}
+	}
+
+	return pgsql.SortExtensions(extensions)
+}
+
+// declaredExtensions reads schema.Metadata["extensions"], which accepts either a list or a
+// comma-separated string. Unknown names are kept: the metadata is an explicit instruction.
+func declaredExtensions(schema *models.Schema) []string {
+	value, ok := schema.Metadata["extensions"]
+	if !ok {
+		return nil
+	}
+
+	var names []string
+	switch declared := value.(type) {
+	case string:
+		names = strings.Split(declared, ",")
+	case []string:
+		names = declared
+	case []any:
+		for _, item := range declared {
+			if name, ok := item.(string); ok {
+				names = append(names, name)
+			}
+		}
+	default:
+		return nil
+	}
+
+	cleaned := make([]string, 0, len(names))
+	for _, name := range names {
+		if name = strings.TrimSpace(name); name != "" {
+			cleaned = append(cleaned, name)
+		}
+	}
+	return cleaned
+}
+
+// btreeCompanionExtension returns btree_gin or btree_gist when a GIN/GiST index covers a
+// scalar type that neither access method has a built-in operator class for. Without the
+// companion extension PostgreSQL rejects the CREATE INDEX outright.
+func btreeCompanionExtension(indexType string, col *models.Column, opClass string) string {
+	if opClass != "" {
+		return ""
+	}
+
+	method := strings.ToLower(strings.TrimSpace(indexType))
+	if method != "gin" && method != "gist" {
+		return ""
+	}
+
+	sqlType := effectiveColumnSQLType(col)
+	if pgsql.IsArrayType(sqlType) {
+		return ""
+	}
+
+	baseType := pgsql.CanonicalizeBaseType(pgsql.ExtractBaseTypeLower(sqlType))
+	if pgsql.TypeExtension(baseType) != "" {
+		// Extension types (geometry, vector, citext, …) ship their own operator classes.
+		return ""
+	}
+
+	if method == "gin" {
+		if nativeGinBaseType(baseType) {
+			return ""
+		}
+		return "btree_gin"
+	}
+	if nativeGistBaseType(baseType) {
+		return ""
+	}
+	return "btree_gist"
+}
+
+// nativeGinBaseType reports whether core PostgreSQL provides a GIN operator class.
+func nativeGinBaseType(baseType string) bool {
+	switch baseType {
+	case "jsonb", "json", "tsvector", "tsquery":
+		return true
+	default:
+		return false
+	}
+}
+
+// nativeGistBaseType reports whether core PostgreSQL provides a GiST operator class.
+func nativeGistBaseType(baseType string) bool {
+	switch baseType {
+	case "tsvector", "tsquery", "point", "box", "circle", "polygon", "line", "lseg", "path", "inet", "cidr":
 		return true
 	}
-			}
-		}
-	}
-	return false
+	return strings.HasSuffix(baseType, "range") || strings.HasSuffix(baseType, "multirange")
 }
 
 func resolveIndexColumn(table *models.Table, colName string) (*models.Column, bool) {
@@ -1642,14 +1886,21 @@ func formatStringList(items []string) string {
 
 // extractOperatorClass extracts operator class from index comment/note
 // Looks for common operator classes like gin_trgm_ops, gist_trgm_ops, etc.
+// explicitOperatorClassPattern matches an "opclass=<name>" hint, the form the PostgreSQL
+// reader uses to carry an index's operator class through the model.
+var explicitOperatorClassPattern = regexp.MustCompile(`(?i)\bopclass\s*=\s*([a-z_][a-z0-9_]*)\b`)
+
 func extractOperatorClass(comment string) string {
 	if comment == "" {
 		return ""
 	}
+
 	lowerComment := strings.ToLower(comment)
-	// Common GIN/GiST operator classes
-	opClasses := []string{"gin_trgm_ops", "gist_trgm_ops", "gin_bigm_ops", "jsonb_ops", "jsonb_path_ops", "array_ops"}
-	for _, op := range opClasses {
+	if matches := explicitOperatorClassPattern.FindStringSubmatch(lowerComment); len(matches) > 1 {
+		return matches[1]
+	}
+
+	for _, op := range knownOperatorClasses() {
 		if strings.Contains(lowerComment, op) {
 			return op
 		}
@@ -1657,6 +1908,35 @@ func extractOperatorClass(comment string) string {
 	return ""
 }
 
+// knownOperatorClasses lists every operator class recognized in an index comment,
+// longest name first so that e.g. gist_geometry_ops_nd wins over a shorter prefix.
+var knownOperatorClasses = sync.OnceValue(func() []string {
+	names := []string{"gin_trgm_ops", "gist_trgm_ops", "gin_bigm_ops", "jsonb_ops", "jsonb_path_ops", "array_ops"}
+	for name := range vectorOperatorClasses {
+		names = append(names, name)
+	}
+	for name := range spatialOperatorClasses {
+		names = append(names, name)
+	}
+	sort.Slice(names, func(i, j int) bool {
+		if len(names[i]) != len(names[j]) {
+			return len(names[i]) > len(names[j])
+		}
+		return names[i] < names[j]
+	})
+	return names
+})
+
+// indexStorageParameters extracts access-method storage parameters from an index comment.
+// Only well-formed "key = value" pairs are kept, so comment prose cannot leak into DDL.
+// Example: "opclass=vector_cosine_ops with (m=16, ef_construction=64)" -> "m = 16, ef_construction = 64".
+func indexStorageParameters(comment string) string {
+	if comment == "" {
+		return ""
+	}
+	return pgsql.FormatStorageParameters(pgsql.ExtractWithClause(comment))
+}
+
 // escapeQuote escapes single quotes in strings for SQL
 func escapeQuote(s string) string {
 	return strings.ReplaceAll(s, "'", "''")

pkg/writers/pgsql/writer_test.go

@@ -1310,3 +1310,203 @@ func TestWriteSchema_UsesStorageTypeForSerialAlterStatements(t *testing.T) {
 		t.Fatalf("expected serial alter to include USING cast, got:\n%s", output)
 	}
 }
+
+// buildVectorSpatialSchema returns a database with a pgvector column and a PostGIS column.
+func buildVectorSpatialSchema(indexType, indexComment string) *models.Database {
+	db := models.InitDatabase("testdb")
+	schema := models.InitSchema("public")
+
+	table := models.InitTable("documents", "public")
+
+	embedding := models.InitColumn("embedding", "documents", "public")
+	embedding.Type = "vector(1536)"
+	table.Columns["embedding"] = embedding
+
+	location := models.InitColumn("location", "documents", "public")
+	location.Type = "geometry(Point,4326)"
+	table.Columns["location"] = location
+
+	if indexType != "" {
+		index := &models.Index{
+			Name:    "idx_documents_embedding",
+			Type:    indexType,
+			Columns: []string{"embedding"},
+			Comment: indexComment,
+		}
+		table.Indexes[index.Name] = index
+	}
+
+	schema.Tables = append(schema.Tables, table)
+	db.Schemas = append(db.Schemas, schema)
+	return db
+}
+
+func writeDatabaseOutput(t *testing.T, db *models.Database) string {
+	t.Helper()
+
+	var buf bytes.Buffer
+	writer := NewWriter(&writers.WriterOptions{})
+	writer.writer = &buf
+
+	if err := writer.WriteDatabase(db); err != nil {
+		t.Fatalf("WriteDatabase failed: %v", err)
+	}
+	return buf.String()
+}
+
+func TestWriteDatabase_VectorAndPostGISColumnsCreateExtensions(t *testing.T) {
+	output := writeDatabaseOutput(t, buildVectorSpatialSchema("", ""))
+
+	for _, want := range []string{
+		"CREATE EXTENSION IF NOT EXISTS postgis;",
+		"CREATE EXTENSION IF NOT EXISTS vector;",
+		"vector(1536)",
+		"geometry(Point,4326)",
+	} {
+		if !strings.Contains(output, want) {
+			t.Fatalf("expected output to contain %q, got:\n%s", want, output)
+		}
+	}
+
+	// postgis must be created before postgis-dependent extensions and stay deterministic
+	if strings.Index(output, "EXISTS postgis;") > strings.Index(output, "EXISTS vector;") {
+		t.Fatalf("expected extensions to be emitted in sorted order, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_HNSWIndexUsesDefaultVectorOperatorClass(t *testing.T) {
+	output := writeDatabaseOutput(t, buildVectorSpatialSchema("hnsw", ""))
+
+	if !strings.Contains(output, "USING hnsw (embedding vector_cosine_ops)") {
+		t.Fatalf("expected hnsw index with default vector operator class, got:\n%s", output)
+	}
+	if !strings.Contains(output, "CREATE EXTENSION IF NOT EXISTS vector;") {
+		t.Fatalf("expected pgvector extension, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_VectorIndexHonoursRequestedOperatorClassAndStorageParameters(t *testing.T) {
+	output := writeDatabaseOutput(t, buildVectorSpatialSchema("ivfflat", "opclass=vector_l2_ops; with (lists=100)"))
+
+	if !strings.Contains(output, "USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)") {
+		t.Fatalf("expected ivfflat index with requested opclass and storage parameters, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_VectorIndexIgnoresIncompatibleOperatorClass(t *testing.T) {
+	output := writeDatabaseOutput(t, buildVectorSpatialSchema("hnsw", "opclass=halfvec_l2_ops"))
+
+	if !strings.Contains(output, "USING hnsw (embedding vector_cosine_ops)") {
+		t.Fatalf("expected halfvec operator class to be rejected for a vector column, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_VectorIndexIgnoresCommentProseInStorageParameters(t *testing.T) {
+	output := writeDatabaseOutput(t, buildVectorSpatialSchema("hnsw", "tuned with (m=16, ef_construction=64, drop table foo)"))
+
+	if !strings.Contains(output, "WITH (m = 16, ef_construction = 64)") {
+		t.Fatalf("expected only well-formed storage parameters, got:\n%s", output)
+	}
+	if strings.Contains(output, "drop table") {
+		t.Fatalf("expected prose to be dropped from storage parameters, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_GistIndexOnGeometryUsesDefaultOperatorClass(t *testing.T) {
+	db := models.InitDatabase("testdb")
+	schema := models.InitSchema("public")
+
+	table := models.InitTable("places", "public")
+	geom := models.InitColumn("geom", "places", "public")
+	geom.Type = "geometry(Point,4326)"
+	table.Columns["geom"] = geom
+
+	table.Indexes["idx_places_geom"] = &models.Index{
+		Name:    "idx_places_geom",
+		Type:    "gist",
+		Columns: []string{"geom"},
+	}
+
+	schema.Tables = append(schema.Tables, table)
+	db.Schemas = append(db.Schemas, schema)
+
+	output := writeDatabaseOutput(t, db)
+
+	if !strings.Contains(output, "USING gist (geom)") {
+		t.Fatalf("expected gist index to rely on the PostGIS default operator class, got:\n%s", output)
+	}
+}
+
+func TestWriteDatabase_GistIndexHonoursRequestedSpatialOperatorClass(t *testing.T) {
+	db := models.InitDatabase("testdb")
+	schema := models.InitSchema("public")
+
+	table := models.InitTable("places", "public")
+	geom := models.InitColumn("geom", "places", "public")
+	geom.Type = "geometry(PointZ,4326)"
+	table.Columns["geom"] = geom
+
+	table.Indexes["idx_places_geom_nd"] = &models.Index{
+		Name:    "idx_places_geom_nd",
+		Type:    "gist",
+		Columns: []string{"geom"},
+		Comment: "opclass=gist_geometry_ops_nd",
+	}
+
+	schema.Tables = append(schema.Tables, table)
+	db.Schemas = append(db.Schemas, schema)
+
+	output := writeDatabaseOutput(t, db)
+
+	if !strings.Contains(output, "USING gist (geom gist_geometry_ops_nd)") {
+		t.Fatalf("expected requested spatial operator class, got:\n%s", output)
+	}
+}
+
+func TestGenerateDatabaseStatements_VectorIndexIncludesOperatorClassAndParameters(t *testing.T) {
+	db := buildVectorSpatialSchema("hnsw", "opclass=vector_ip_ops; with (m=16)")
+
+	writer := NewWriter(&writers.WriterOptions{})
+	statements, err := writer.GenerateDatabaseStatements(db)
+	if err != nil {
+		t.Fatalf("GenerateDatabaseStatements failed: %v", err)
+	}
+
+	joined := strings.Join(statements, "\n")
+	for _, want := range []string{
+		"CREATE EXTENSION IF NOT EXISTS vector",
+		"CREATE EXTENSION IF NOT EXISTS postgis",
+		"USING hnsw (embedding vector_ip_ops) WITH (m = 16)",
+	} {
+		if !strings.Contains(joined, want) {
+			t.Fatalf("expected statements to contain %q, got:\n%s", want, joined)
+		}
+	}
+}
+
+func TestIndexStorageParameters(t *testing.T) {
+	tests := []struct {
+		name    string
+		comment string
+		want    string
+	}{
+		{"empty", "", ""},
+		{"no with clause", "opclass=vector_cosine_ops", ""},
+		{"single parameter", "with (lists=100)", "lists = 100"},
+		{"multiple parameters", "WITH (m = 16, ef_construction = 64)", "m = 16, ef_construction = 64"},
+		{"quoted value kept", "with (fillfactor='90')", "fillfactor = '90'"},
+		{"bm25 key field", "with (key_field='id')", "key_field = 'id'"},
+		{"dollar quoted value", "with (options = $$[build.internal]\nlists = [4096]$$)", "options = $$[build.internal]\nlists = [4096]$$"},
+		{"dollar quoted value with parens", "with (options = $$f(x)$$, m = 16)", "options = $$f(x)$$, m = 16"},
+		{"prose dropped", "with (lists=100, please drop everything)", "lists = 100"},
+		{"unterminated quote dropped", "with (key_field='id)", ""},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := indexStorageParameters(tt.comment); got != tt.want {
+				t.Errorf("indexStorageParameters(%q) = %q, want %q", tt.comment, got, tt.want)
+			}
+		})
+	}
+}

pkg/writers/prisma/writer.go

@@ -27,7 +27,7 @@ func (w *Writer) WriteDatabase(db *models.Database) error {
 	content := w.databaseToPrisma(db)
 
 	if w.options.OutputPath != "" {
-		return os.WriteFile(w.options.OutputPath, []byte(content), 0644)
+		return os.WriteFile(w.options.OutputPath, []byte(content), 0o644)
 	}
 
 	fmt.Print(content)

pkg/writers/sqlite/README.md

@@ -118,6 +118,21 @@ CREATE TABLE "posts" (
 - **Check Constraints**: Generated as comments (should be added to CREATE TABLE manually)
 - **Indexes**: Generated without PostgreSQL-specific features (no GIN, GiST, operator classes)
 
+## DBML dialect directives
+
+`@sqlite:` directives carried on a model object's `Metadata` (typically from a
+DBML source file) are translated to SQL:
+
+| Directive | Location | Emitted |
+|-----------|----------|---------|
+| `@sqlite: without rowid` | table | `WITHOUT ROWID` table option |
+| `@sqlite: strict` | table | `STRICT` table option (after `WITHOUT ROWID`) |
+| `@sqlite(col): collate …` | column | ` COLLATE …` in the column definition |
+
+Directives for other dialects (`@postgres:` …) are ignored. With
+`WriterOptions.StrictDirectives` (CLI `--strict-directives`) an untranslatable
+`@sqlite:` key is an error. Full reference: [`docs/DBML_DIRECTIVES.md`](../../../docs/DBML_DIRECTIVES.md).
+
 ## Output Structure
 
 Generated SQL follows this order:

pkg/writers/sqlite/directives.go

@@ -0,0 +1,84 @@
+package sqlite
+
+import (
+	"fmt"
+	"strings"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+)
+
+// directiveNamespace is the dialect namespace this writer consumes. Directives
+// for other namespaces (e.g. "postgres") are ignored and never emitted as SQL.
+const directiveNamespace = "sqlite"
+
+// sqliteHandledDirectives maps a directive location to the set of sqlite keys
+// this writer knows how to translate. In strict mode an unknown key for this
+// namespace at a supported location is a hard error.
+var sqliteHandledDirectives = map[string]map[string]bool{
+	models.DirectiveLocationTable:  {"without": true, "strict": true},
+	models.DirectiveLocationColumn: {"collate": true},
+}
+
+// checkDirectives validates sqlite directives across a schema when strict mode
+// is enabled. With strict mode off it is a no-op.
+func (w *Writer) checkDirectives(schema *models.Schema) error {
+	if w.options == nil || !w.options.StrictDirectives {
+		return nil
+	}
+	for _, table := range schema.Tables {
+		if err := checkObjectDirectives(table.Metadata, models.DirectiveLocationTable, table.Name); err != nil {
+			return err
+		}
+		for _, col := range table.Columns {
+			if err := checkObjectDirectives(col.Metadata, models.DirectiveLocationColumn, table.Name+"."+col.Name); err != nil {
+				return err
+			}
+		}
+		for _, idx := range table.Indexes {
+			if err := checkObjectDirectives(idx.Metadata, models.DirectiveLocationIndex, idx.Name); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+func checkObjectDirectives(meta map[string]any, location, owner string) error {
+	for _, d := range models.DirectivesForNamespace(meta, directiveNamespace) {
+		if !sqliteHandledDirectives[location][d.Key] {
+			return fmt.Errorf("sqlite: %s: unsupported @sqlite directive %q at %s level (strict mode)", owner, d.Key, location)
+		}
+	}
+	return nil
+}
+
+// sqliteTableOptions returns the trailing table-option clause for a CREATE TABLE
+// statement, e.g. "WITHOUT ROWID, STRICT". WITHOUT ROWID is emitted before
+// STRICT, matching SQLite's own grammar ordering.
+func sqliteTableOptions(table *models.Table) string {
+	var opts []string
+	if models.HasDirective(table.Metadata, directiveNamespace, "without") {
+		opts = append(opts, "WITHOUT ROWID")
+	}
+	if models.HasDirective(table.Metadata, directiveNamespace, "strict") {
+		opts = append(opts, "STRICT")
+	}
+	return strings.Join(opts, ", ")
+}
+
+// sqliteColumnCollate returns a " COLLATE <name>" clause for a column carrying an
+// @sqlite(col): collate <name> directive, or "".
+func sqliteColumnCollate(col *models.Column) string {
+	for _, d := range models.DirectivesForNamespace(col.Metadata, directiveNamespace) {
+		if d.Key != "collate" {
+			continue
+		}
+		name := strings.TrimSpace(strings.TrimPrefix(strings.TrimSpace(d.Args), "collate"))
+		name = strings.TrimSpace(name)
+		if name == "" {
+			return ""
+		}
+		return " COLLATE " + name
+	}
+	return ""
+}

pkg/writers/sqlite/directives_test.go

@@ -0,0 +1,82 @@
+package sqlite
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"git.warky.dev/wdevs/relspecgo/pkg/models"
+	"git.warky.dev/wdevs/relspecgo/pkg/writers"
+)
+
+func sqliteDirectiveDB(t *testing.T) *models.Database {
+	t.Helper()
+	db := models.InitDatabase("testdb")
+	schema := models.InitSchema("public")
+	table := models.InitTable("events", "public")
+
+	id := models.InitColumn("id", "events", "public")
+	id.Type = "bigint"
+	id.IsPrimaryKey = true
+	id.NotNull = true
+	table.Columns["id"] = id
+
+	name := models.InitColumn("name", "events", "public")
+	name.Type = "varchar(200)"
+	name.NotNull = true
+	models.AddDirective(name.Metadata, models.Directive{Namespace: "sqlite", Args: "collate NOCASE"})
+	// A postgres directive on the same column must be ignored by the sqlite writer.
+	models.AddDirective(name.Metadata, models.Directive{Namespace: "postgres", Args: "storage plain"})
+	table.Columns["name"] = name
+
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "sqlite", Args: "without rowid"})
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "sqlite", Args: "strict"})
+	models.AddDirective(table.Metadata, models.Directive{Namespace: "postgres", Args: "partition by RANGE (id)"})
+
+	schema.Tables = append(schema.Tables, table)
+	db.Schemas = append(db.Schemas, schema)
+	return db
+}
+
+func TestSqliteDirectives_TableOptionsAndCollate(t *testing.T) {
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{})
+	w.writer = &buf
+	if err := w.WriteDatabase(sqliteDirectiveDB(t)); err != nil {
+		t.Fatalf("WriteDatabase: %v", err)
+	}
+	out := buf.String()
+
+	if !strings.Contains(out, ") WITHOUT ROWID, STRICT;") {
+		t.Errorf("missing table options clause:\n%s", out)
+	}
+	if !strings.Contains(out, `"name" TEXT COLLATE NOCASE NOT NULL`) {
+		t.Errorf("missing column COLLATE clause:\n%s", out)
+	}
+	// postgres directives must never reach sqlite output.
+	if strings.Contains(strings.ToUpper(out), "PARTITION BY") || strings.Contains(strings.ToUpper(out), "STORAGE PLAIN") {
+		t.Errorf("postgres directive leaked into sqlite output:\n%s", out)
+	}
+}
+
+func TestSqliteDirectives_StrictUnknownKeyErrors(t *testing.T) {
+	db := sqliteDirectiveDB(t)
+	models.AddDirective(db.Schemas[0].Tables[0].Metadata, models.Directive{Namespace: "sqlite", Args: "frobnicate x"})
+
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{StrictDirectives: true})
+	w.writer = &buf
+	err := w.WriteDatabase(db)
+	if err == nil || !strings.Contains(err.Error(), "frobnicate") {
+		t.Fatalf("want strict error for unknown sqlite key, got %v", err)
+	}
+}
+
+func TestSqliteDirectives_StrictIgnoresPostgres(t *testing.T) {
+	var buf bytes.Buffer
+	w := NewWriter(&writers.WriterOptions{StrictDirectives: true})
+	w.writer = &buf
+	if err := w.WriteDatabase(sqliteDirectiveDB(t)); err != nil {
+		t.Fatalf("strict mode should ignore postgres directives, got %v", err)
+	}
+}

pkg/writers/sqlite/template_functions.go

@@ -25,6 +25,7 @@ func GetTemplateFuncs(opts *writers.WriterOptions) template.FuncMap {
 		"join":           strings.Join,
 		"lower":          strings.ToLower,
 		"upper":          strings.ToUpper,
+		"column_collate": sqliteColumnCollate,
 	}
 }