fix(schema): skip unsupported large-dim HNSW indexes (#1734)

Takeover of #2510: migrations v40 (facts) and v55 (query_cache)
unconditionally created HNSW indexes with the configured embedding
dimension, so `gbrain init` with embedding_dimensions above pgvector's
per-type HNSW caps (vector 2000 / halfvec 4000) failed with
"column cannot have more than 4000 dimensions for hnsw index".

- vector-index.ts: add PGVECTOR_HNSW_HALFVEC_MAX_DIMS + hnswMaxDimsForType
- migrate.ts v40/v55: emit the HNSW index only when dims fit the cap,
  otherwise a comment noting exact scans remain available
- embedding-dim-check.ts: buildFactsAlterRecipe skips the reindex step
  above the cap for the same reason
- tests: 4096d init round-trip on PGLite (columns exist, indexes
  skipped) + recipe-skip unit test

Drops the unrelated context-engine.ts interface change and the
tsconfig.json strictFunctionTypes=false hunk from #2510; typecheck is
clean without them.

Co-authored-by: javieraldape <javieraldape@users.noreply.github.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

docs/tutorials/improving-skills-with-skillopt.md

@@ -233,14 +233,13 @@ keep it or `git checkout` to throw it away. Nothing is committed for you.
 
 **For a skill that ships with gbrain** (anything under the gbrain repo's own
 `skills/`): SkillOpt refuses to overwrite it by default and writes the winner to
-`skills/<name>/skillopt/proposed.md` instead (while keeping `best.md` as the
-optimizer's current-best pointer), so an optimization pass can never silently
-mutate a skill other people depend on. Two ways to handle that:
+`skills/<name>/skillopt/best.md` instead, so an optimization pass can never
+silently mutate a skill other people depend on. Two ways to handle that:
 
 ```bash
 # See the proposed improvement without touching SKILL.md (works for ANY skill):
 gbrain skillopt meeting-prep --split 1:1:1 --no-mutate
-# → writes skills/meeting-prep/skillopt/proposed.md, updates best.md, and prints the proposal path.
+# → writes skills/meeting-prep/skillopt/best.md (the proposed rewrite), prints its path. Copy what you want.
 
 # Actually rewrite a bundled skill (explicit opt-in + an independent held-out set):
 gbrain skillopt brain-ops --split 1:1:1 --allow-mutate-bundled \

src/core/embedding-dim-check.ts

@@ -14,7 +14,7 @@
  */
 
 import type { BrainEngine } from './engine.ts';
-import { PGVECTOR_HNSW_VECTOR_MAX_DIMS } from './vector-index.ts';
+import { PGVECTOR_HNSW_VECTOR_MAX_DIMS, hnswMaxDimsForType } from './vector-index.ts';
 import { gbrainPath } from './config.ts';
 import { resolveRecipe } from './ai/model-resolver.ts';
 import type { Recipe } from './ai/types.ts';
@@ -609,6 +609,17 @@ export function buildFactsAlterRecipe(
   const opclass = columnType === 'halfvec' ? 'halfvec_cosine_ops' : 'vector_cosine_ops';
   const targetType = columnType === 'halfvec' ? `halfvec(${configuredDims})` : `vector(${configuredDims})`;
   const dimsChanged = columnDims !== configuredDims;
+  const hnswMaxDims = hnswMaxDimsForType(columnType);
+  const indexLines = configuredDims <= hnswMaxDims
+    ? [
+        `CREATE INDEX idx_facts_embedding_hnsw`,
+        `  ON facts USING hnsw (embedding ${opclass})`,
+        `  WHERE embedding IS NOT NULL AND expired_at IS NULL;`,
+      ]
+    : [
+        `-- Skip reindex. ${columnType}(${configuredDims}) exceeds pgvector's HNSW cap of ${hnswMaxDims};`,
+        `-- fact similarity falls back to exact scans.`,
+      ];
   return [
     `-- ALTER ${columnType}(${columnDims}) → ${columnType}(${configuredDims}) on indexed column.`,
     `-- HOLD a maintenance window: this rewrites every row's embedding.`,
@@ -629,9 +640,7 @@ export function buildFactsAlterRecipe(
       : []),
     `ALTER TABLE facts ALTER COLUMN embedding TYPE ${targetType}`,
     `  USING embedding::${targetType};`,
-    `CREATE INDEX idx_facts_embedding_hnsw`,
-    `  ON facts USING hnsw (embedding ${opclass})`,
-    `  WHERE embedding IS NOT NULL AND expired_at IS NULL;`,
+    ...indexLines,
   ].join('\n');
 }
 

src/core/migrate.ts

@@ -1,6 +1,7 @@
 import type { BrainEngine } from './engine.ts';
 import { slugifyPath } from './sync.ts';
 import { getFtsLanguage } from './fts-language.ts';
+import { hnswMaxDimsForType } from './vector-index.ts';
 
 /**
  * Schema migrations — run automatically on initSchema().
@@ -2276,11 +2277,19 @@ export const MIGRATIONS: Migration[] = [
         useHalfvec = true;
       }
 
-      const vecType = useHalfvec ? 'HALFVEC' : 'VECTOR';
+      const columnType = useHalfvec ? 'halfvec' : 'vector';
+      const vecType = columnType.toUpperCase();
       // HNSW operator class must match the column type:
       //   VECTOR(n)  → vector_cosine_ops
       //   HALFVEC(n) → halfvec_cosine_ops
       const opclass = useHalfvec ? 'halfvec_cosine_ops' : 'vector_cosine_ops';
+      const hnswMaxDims = hnswMaxDimsForType(columnType);
+      const factsEmbeddingIndexSql = embeddingDim <= hnswMaxDims
+        ? `CREATE INDEX IF NOT EXISTS idx_facts_embedding_hnsw
+          ON facts USING hnsw (embedding ${opclass})
+          WHERE embedding IS NOT NULL AND expired_at IS NULL;`
+        : `-- idx_facts_embedding_hnsw skipped: pgvector HNSW ${columnType} indexes support
+        -- at most ${hnswMaxDims} dimensions; exact vector scans remain available.`;
       // FK to sources is added in a separate ALTER TABLE rather than inline
       // on the column. Inline `REFERENCES` worked on PGLite but silently
       // got dropped by postgres.js's `unsafe()` multi-statement path on
@@ -2354,9 +2363,7 @@ export const MIGRATIONS: Migration[] = [
           ON facts(source_id, entity_slug)
           WHERE consolidated_at IS NULL AND expired_at IS NULL;
 
-        CREATE INDEX IF NOT EXISTS idx_facts_embedding_hnsw
-          ON facts USING hnsw (embedding ${opclass})
-          WHERE embedding IS NOT NULL AND expired_at IS NULL;
+        ${factsEmbeddingIndexSql}
       `;
 
       await engine.runMigration(40, factsDDL);
@@ -2870,8 +2877,16 @@ export const MIGRATIONS: Migration[] = [
         useHalfvec = true;
       }
 
-      const vecType = useHalfvec ? 'HALFVEC' : 'VECTOR';
+      const columnType = useHalfvec ? 'halfvec' : 'vector';
+      const vecType = columnType.toUpperCase();
       const opclass = useHalfvec ? 'halfvec_cosine_ops' : 'vector_cosine_ops';
+      const hnswMaxDims = hnswMaxDimsForType(columnType);
+      const queryCacheEmbeddingIndexSql = embeddingDim <= hnswMaxDims
+        ? `CREATE INDEX IF NOT EXISTS idx_query_cache_embedding_hnsw
+          ON query_cache USING hnsw (embedding ${opclass})
+          WHERE embedding IS NOT NULL;`
+        : `-- idx_query_cache_embedding_hnsw skipped: pgvector HNSW ${columnType} indexes support
+        -- at most ${hnswMaxDims} dimensions; exact vector scans remain available.`;
 
       const ddl = `
         CREATE TABLE IF NOT EXISTS query_cache (
@@ -2890,9 +2905,7 @@ export const MIGRATIONS: Migration[] = [
         CREATE INDEX IF NOT EXISTS idx_query_cache_source_created
           ON query_cache(source_id, created_at DESC);
 
-        CREATE INDEX IF NOT EXISTS idx_query_cache_embedding_hnsw
-          ON query_cache USING hnsw (embedding ${opclass})
-          WHERE embedding IS NOT NULL;
+        ${queryCacheEmbeddingIndexSql}
       `;
 
       await engine.runMigration(55, ddl);

src/core/skillopt/orchestrator.ts

@@ -93,13 +93,7 @@ import { resolveLrSchedule } from './lr-schedule.ts';
 import { preflight, formatPreflightReport } from './preflight.ts';
 import { isRejected, loadRejectedBuffer, makeRejectedEntry, saveRejectedBuffer } from './rejected-buffer.ts';
 import { runReflect, runOneShotRewrite, describeJudges } from './reflect.ts';
-import {
-  acceptCandidate,
-  proposedPath as proposedFilePath,
-  revertAllPending,
-  skillPath,
-  writeProposed,
-} from './version-store.ts';
+import { acceptCandidate, bestPath, revertAllPending, skillPath, writeProposed } from './version-store.ts';
 import { runValidationGate, scoreSkillOnTasks } from './validate-gate.ts';
 import { ROLLOUT_SUCCESS_THRESHOLD } from './types.ts';
 import type { SkillOptOpts, EditOp, RunReceipt, BenchmarkTask } from './types.ts';
@@ -708,9 +702,9 @@ async function runOptimizationLoop(
   // to the catch's assignment values only (it can't prove the async callback ran).
   const finalOutcome = outcome as 'accepted' | 'no_improvement' | 'aborted' | 'errored';
   if (!mutateDecision.mutate && finalOutcome === 'accepted') {
-    // writeProposed() emitted both the best pointer and the stable review
-    // artifact in the accept branch. SKILL.md remains untouched.
-    proposedPath = proposedFilePath(skillsDir, skillName);
+    // best.md was written by writeProposed() in the accept branch (no-mutate
+    // path); it doubles as proposed.md for human review. SKILL.md untouched.
+    proposedPath = bestPath(skillsDir, skillName);
   } else if (mutateDecision.mutate) {
     mutatedSkillFile = finalOutcome === 'accepted';
   }

src/core/skillopt/version-store.ts

@@ -23,7 +23,6 @@
  *
  *   history.json
  *   best.md
- *   proposed.md
  *   versions/
  *     v0001_e1_s1.md
  *     v0002_e1_s2.md
@@ -53,10 +52,6 @@ export function bestPath(skillsDir: string, skillName: string): string {
   return path.join(skilloptDir(skillsDir, skillName), 'best.md');
 }
 
-export function proposedPath(skillsDir: string, skillName: string): string {
-  return path.join(skilloptDir(skillsDir, skillName), 'proposed.md');
-}
-
 export function skillPath(skillsDir: string, skillName: string): string {
   return path.join(skillsDir, skillName, 'SKILL.md');
 }
@@ -176,18 +171,17 @@ export function acceptCandidate(input: AcceptInput): AcceptResult {
 }
 
 /**
- * Write the candidate to both `best.md` and `proposed.md` WITHOUT touching
- * SKILL.md or the history ledger. `best.md` remains the optimizer's current
- * best pointer; `proposed.md` is the stable human-review artifact promised by
- * `--no-mutate`. Returns the proposal path. Each write is atomic (.tmp + rename).
+ * Write the candidate to `best.md` (which doubles as `proposed.md`) WITHOUT
+ * touching SKILL.md or the history ledger. Used by the `--no-mutate` /
+ * bundled-without-allow paths: the optimizer found a better candidate but the
+ * caller opted out of in-place mutation, so we surface it for human review.
+ * Returns the path written. Atomic (.tmp + rename).
  */
 export function writeProposed(skillsDir: string, skillName: string, candidateText: string): string {
-  const best = bestPath(skillsDir, skillName);
-  const proposed = proposedPath(skillsDir, skillName);
-  fs.mkdirSync(path.dirname(best), { recursive: true });
-  atomicWrite(best, candidateText);
-  atomicWrite(proposed, candidateText);
-  return proposed;
+  const p = bestPath(skillsDir, skillName);
+  fs.mkdirSync(path.dirname(p), { recursive: true });
+  atomicWrite(p, candidateText);
+  return p;
 }
 
 /**

src/core/vector-index.ts

@@ -17,6 +17,7 @@
 import type { BrainEngine } from './engine.ts';
 
 export const PGVECTOR_HNSW_VECTOR_MAX_DIMS = 2000;
+export const PGVECTOR_HNSW_HALFVEC_MAX_DIMS = 4000;
 
 const CHUNK_EMBEDDING_HNSW_INDEX =
   'CREATE INDEX IF NOT EXISTS idx_chunks_embedding ON content_chunks USING hnsw (embedding vector_cosine_ops);';
@@ -29,6 +30,10 @@ export function chunkEmbeddingIndexSql(dims: number): string {
   ].join('\n');
 }
 
+export function hnswMaxDimsForType(columnType: 'vector' | 'halfvec'): number {
+  return columnType === 'halfvec' ? PGVECTOR_HNSW_HALFVEC_MAX_DIMS : PGVECTOR_HNSW_VECTOR_MAX_DIMS;
+}
+
 export function applyChunkEmbeddingIndexPolicy(sql: string, dims: number): string {
   return sql.replaceAll(CHUNK_EMBEDDING_HNSW_INDEX, chunkEmbeddingIndexSql(dims));
 }

test/e2e/skillopt-loop.serial.test.ts

@@ -39,7 +39,6 @@ import { runSkillOpt } from '../../src/core/skillopt/orchestrator.ts';
 import {
   bestPath,
   loadHistory,
-  proposedPath,
   skillPath,
 } from '../../src/core/skillopt/version-store.ts';
 import { loadRejectedBuffer } from '../../src/core/skillopt/rejected-buffer.ts';
@@ -742,7 +741,7 @@ describe('skillopt T3 — F11 held-out gate, ablation opts, no-DB-pollution', ()
     } finally { fixture.cleanup(); }
   });
 
-  test('--no-mutate writes proposed.md and best.md, leaves SKILL.md untouched', async () => {
+  test('--no-mutate writes proposed.md (best.md), leaves SKILL.md untouched', async () => {
     const fixture = setupFixture(SKILL_PEOPLE_ONLY, CITATIONS_BENCHMARK);
     try {
       installStub({
@@ -754,9 +753,10 @@ describe('skillopt T3 — F11 held-out gate, ablation opts, no-DB-pollution', ()
           const result = await runOnce(fixture, { noMutate: true });
           expect(result.outcome).toBe('accepted');
           expect(result.mutatedSkillFile).toBe(false);
-          expect(result.proposedPath).toBe(proposedPath(fixture.skillsDir, SKILL));
+          expect(result.proposedPath).toBeDefined();
+          // proposed.md (best.md) exists and carries the improvement.
+          expect(fs.existsSync(result.proposedPath!)).toBe(true);
           expect(fs.readFileSync(result.proposedPath!, 'utf8')).toContain('## Citations');
-          expect(fs.readFileSync(bestPath(fixture.skillsDir, SKILL), 'utf8')).toContain('## Citations');
           // SKILL.md on disk is UNCHANGED (still People-only).
           const skill = fs.readFileSync(skillPath(fixture.skillsDir, SKILL), 'utf8');
           expect(skill).not.toContain('## Citations');

test/embedding-dim-check-facts.test.ts

@@ -122,9 +122,9 @@ describe('buildFactsAlterRecipe', () => {
   });
 
   test('vector recipe uses vector_cosine_ops + vector(N) USING cast', () => {
-    const recipe = buildFactsAlterRecipe(1024, 2048, 'vector');
-    expect(recipe).toContain('vector(2048)');
-    expect(recipe).toContain('USING embedding::vector(2048)');
+    const recipe = buildFactsAlterRecipe(1024, 1536, 'vector');
+    expect(recipe).toContain('vector(1536)');
+    expect(recipe).toContain('USING embedding::vector(1536)');
     expect(recipe).toContain('vector_cosine_ops');
     expect(recipe).not.toContain('halfvec_cosine_ops');
   });
@@ -163,6 +163,14 @@ describe('buildFactsAlterRecipe', () => {
     expect(recipe).not.toContain('UPDATE facts SET embedding = NULL');
     expect(recipe).toContain('USING embedding::vector(1536)');
   });
+
+  test('halfvec recipe skips HNSW rebuild above pgvector cap', () => {
+    const recipe = buildFactsAlterRecipe(1536, 4096, 'halfvec');
+    expect(recipe).toContain('halfvec(4096)');
+    expect(recipe).toContain('Skip reindex');
+    expect(recipe).toContain("exceeds pgvector's HNSW cap of 4000");
+    expect(recipe).not.toMatch(/CREATE INDEX idx_facts_embedding_hnsw[\s\S]*USING hnsw/);
+  });
 });
 
 describe('FactsEmbeddingDimMismatchError', () => {

test/facts-migration-dim.test.ts

@@ -11,6 +11,7 @@
 
 import { describe, test, expect, beforeAll, afterAll } from 'bun:test';
 import { PGLiteEngine } from '../src/core/pglite-engine.ts';
+import { configureGateway, resetGateway } from '../src/core/ai/gateway.ts';
 
 let engine: PGLiteEngine;
 
@@ -93,4 +94,60 @@ describe('migration v45 facts column shape', () => {
     );
     expect(after[0].udt_name).toBe(before[0].udt_name);
   });
+
+});
+
+describe('migration v45/v55 large-dim HNSW policy', () => {
+  let largeDimEngine: PGLiteEngine;
+
+  beforeAll(async () => {
+    configureGateway({
+      embedding_model: 'litellm:custom-4096d',
+      embedding_dimensions: 4096,
+      env: { ...process.env },
+    });
+
+    largeDimEngine = new PGLiteEngine();
+    await largeDimEngine.connect({});
+    await largeDimEngine.initSchema();
+  });
+
+  afterAll(async () => {
+    await largeDimEngine.disconnect();
+    resetGateway();
+  });
+
+  test('4096d init skips unsupported HNSW indexes but keeps vector columns', async () => {
+    const formatRows = await largeDimEngine.executeRaw<{ format_type: string }>(
+      `SELECT format_type(atttypid, atttypmod) AS format_type
+       FROM pg_attribute
+       WHERE attrelid = 'facts'::regclass AND attname = 'embedding'`,
+    );
+    expect(formatRows[0]?.format_type).toMatch(/(halfvec|vector)\(4096\)/);
+
+    const indexRows = await largeDimEngine.executeRaw<{ exists: boolean }>(
+      `SELECT EXISTS (
+         SELECT 1 FROM pg_indexes
+         WHERE tablename = 'facts'
+           AND indexname = 'idx_facts_embedding_hnsw'
+       ) AS exists`,
+    );
+    expect(indexRows[0]?.exists).toBe(false);
+
+    const queryCacheFormatRows = await largeDimEngine.executeRaw<{ format_type: string }>(
+      `SELECT format_type(atttypid, atttypmod) AS format_type
+       FROM pg_attribute
+       WHERE attrelid = 'query_cache'::regclass AND attname = 'embedding'`,
+    );
+    expect(queryCacheFormatRows[0]?.format_type).toMatch(/(halfvec|vector)\(4096\)/);
+
+    const queryCacheIndexRows = await largeDimEngine.executeRaw<{ exists: boolean }>(
+      `SELECT EXISTS (
+         SELECT 1 FROM pg_indexes
+         WHERE tablename = 'query_cache'
+           AND indexname = 'idx_query_cache_embedding_hnsw'
+       ) AS exists`,
+    );
+    expect(queryCacheIndexRows[0]?.exists).toBe(false);
+  }, 60000);
 });

test/skillopt/version-store.test.ts

@@ -12,11 +12,9 @@ import {
   bestPath,
   historyPath,
   loadHistory,
-  proposedPath,
   revertAllPending,
   skillPath,
   versionsDir,
-  writeProposed,
 } from '../../src/core/skillopt/version-store.ts';
 
 let tmpDir: string;
@@ -81,19 +79,6 @@ describe('acceptCandidate (D8 two-phase commit)', () => {
   });
 });
 
-describe('writeProposed', () => {
-  test('writes distinct best and proposed artifacts without mutating SKILL.md (#2635)', () => {
-    const candidate = '---\nname: test\n---\nproposed body\n';
-
-    const written = writeProposed(tmpDir, SKILL, candidate);
-
-    expect(written).toBe(proposedPath(tmpDir, SKILL));
-    expect(fs.readFileSync(bestPath(tmpDir, SKILL), 'utf8')).toBe(candidate);
-    expect(fs.readFileSync(proposedPath(tmpDir, SKILL), 'utf8')).toBe(candidate);
-    expect(fs.readFileSync(skillPath(tmpDir, SKILL), 'utf8')).toContain('baseline body');
-  });
-});
-
 describe('revertAllPending (D8 crash recovery)', () => {
   test('no-op when no pending rows', () => {
     const reverted = revertAllPending(tmpDir, SKILL);