fix(pages): default chunker_version to MARKDOWN_CHUNKER_VERSION on INSERT (#2807) (#2988)

putPage's INSERT used COALESCE(<chunkerVersion>, 1), so callers that don't
supply chunker_version (no MCP/subagent caller does — it's internal metadata)
landed new pages at version 1. Dream subagents write through putPage directly,
so their pages got v1 and doctor's contextual_retrieval_coverage check flagged
them as "older chunker_version" forever, even though they were chunked and
embedded with the current chunker.

Default the INSERT to MARKDOWN_CHUNKER_VERSION on both engines. The ON CONFLICT
UPDATE still COALESCE-preserves an explicitly supplied version. Add an
engine-level regression test.

src/core/pglite-engine.ts

@@ -23,6 +23,7 @@ import { runMigrations } from './migrate.ts';
 import { PGLITE_SCHEMA_SQL, getPGLiteSchema } from './pglite-schema.ts';
 import { DEFAULT_EMBEDDING_MODEL, DEFAULT_EMBEDDING_DIMENSIONS } from './ai/defaults.ts';
 import { DELETE_BATCH_SIZE } from './engine-constants.ts';
+import { MARKDOWN_CHUNKER_VERSION } from './chunkers/recursive.ts';
 import { acquireLock, releaseLock, type LockHandle } from './pglite-lock.ts';
 import { getFtsLanguage } from './fts-language.ts';
 import type {
@@ -1035,7 +1036,7 @@ export class PGLiteEngine implements BrainEngine {
     const ingestedAt = (sourceKind || sourceUri || ingestedVia) ? new Date().toISOString() : null;
     const { rows } = await this.db.query(
       `INSERT INTO pages (source_id, slug, type, page_kind, title, compiled_truth, timeline, frontmatter, content_hash, updated_at, effective_date, effective_date_source, import_filename, chunker_version, source_path, source_kind, source_uri, ingested_via, ingested_at)
-       VALUES ($1, $2, $3, $4, $5, $6, $7, $8::jsonb, $9, now(), $10::timestamptz, $11, $12, COALESCE($13, 1), $14, $15, $16, $17, $18::timestamptz)
+       VALUES ($1, $2, $3, $4, $5, $6, $7, $8::jsonb, $9, now(), $10::timestamptz, $11, $12, COALESCE($13, ${MARKDOWN_CHUNKER_VERSION}), $14, $15, $16, $17, $18::timestamptz)
        ON CONFLICT (source_id, slug) DO UPDATE SET
          type = EXCLUDED.type,
          page_kind = EXCLUDED.page_kind,

src/core/postgres-engine.ts

@@ -35,6 +35,7 @@ import {
   EmbeddingColumnNotRegisteredError,
 } from './search/embedding-column.ts';
 import { getFtsLanguage } from './fts-language.ts';
+import { MARKDOWN_CHUNKER_VERSION } from './chunkers/recursive.ts';
 import type {
   Page, PageInput, PageFilters, PageType,
   Chunk, ChunkInput, StaleChunkRow, StalePageRow,
@@ -1097,7 +1098,7 @@ export class PostgresEngine implements BrainEngine {
     const ingestedAt = (sourceKind || sourceUri || ingestedVia) ? new Date() : null;
     const rows = await sql`
       INSERT INTO pages (source_id, slug, type, page_kind, title, compiled_truth, timeline, frontmatter, content_hash, updated_at, effective_date, effective_date_source, import_filename, chunker_version, source_path, source_kind, source_uri, ingested_via, ingested_at)
-      VALUES (${sourceId}, ${slug}, ${page.type}, ${pageKind}, ${page.title}, ${page.compiled_truth}, ${page.timeline || ''}, ${sql.json(frontmatter as Parameters<typeof sql.json>[0])}, ${hash}, now(), ${effectiveDate}, ${effectiveDateSource}, ${importFilename}, COALESCE(${chunkerVersion}::smallint, 1), ${sourcePath}, ${sourceKind}, ${sourceUri}, ${ingestedVia}, ${ingestedAt})
+      VALUES (${sourceId}, ${slug}, ${page.type}, ${pageKind}, ${page.title}, ${page.compiled_truth}, ${page.timeline || ''}, ${sql.json(frontmatter as Parameters<typeof sql.json>[0])}, ${hash}, now(), ${effectiveDate}, ${effectiveDateSource}, ${importFilename}, COALESCE(${chunkerVersion}::smallint, ${MARKDOWN_CHUNKER_VERSION}), ${sourcePath}, ${sourceKind}, ${sourceUri}, ${ingestedVia}, ${ingestedAt})
       ON CONFLICT (source_id, slug) DO UPDATE SET
         type = EXCLUDED.type,
         page_kind = EXCLUDED.page_kind,

test/chunker-version-insert-default.test.ts

@@ -0,0 +1,80 @@
+/**
+ * #2807 — putPage INSERT default for chunker_version.
+ *
+ * postgres-engine / pglite-engine used `COALESCE(<chunkerVersion>, 1)` in the
+ * pages INSERT. Callers that don't supply chunker_version (no MCP/subagent
+ * caller does — it's internal metadata) landed new pages at version 1, so
+ * doctor's contextual_retrieval_coverage check flagged dream-written pages as
+ * "older chunker_version" forever, even though they were chunked/embedded with
+ * the current chunker.
+ *
+ * Fix: default the INSERT to MARKDOWN_CHUNKER_VERSION. The ON CONFLICT UPDATE
+ * still COALESCE-preserves an explicitly supplied version.
+ */
+
+import { describe, test, expect, beforeAll, afterAll } from 'bun:test';
+import { PGLiteEngine } from '../src/core/pglite-engine.ts';
+import { MARKDOWN_CHUNKER_VERSION } from '../src/core/chunkers/recursive.ts';
+
+describe('#2807 — putPage chunker_version INSERT default', () => {
+  let engine: PGLiteEngine;
+
+  beforeAll(async () => {
+    engine = new PGLiteEngine();
+    await engine.connect({});
+    await engine.initSchema();
+  });
+
+  afterAll(async () => {
+    await engine.disconnect();
+  }, 30_000);
+
+  async function readChunkerVersion(slug: string): Promise<number> {
+    const { rows } = await (engine as any).db.query(
+      `SELECT chunker_version FROM pages WHERE slug = $1`,
+      [slug],
+    );
+    return Number((rows[0] as { chunker_version: number }).chunker_version);
+  }
+
+  test('page written without chunker_version defaults to MARKDOWN_CHUNKER_VERSION, not 1', async () => {
+    await engine.putPage('dream/no-chunker-version', {
+      type: 'concept',
+      title: 'Dream synthesized page',
+      compiled_truth: 'Written directly through putPage, like a dream subagent.',
+      timeline: '',
+    });
+    expect(await readChunkerVersion('dream/no-chunker-version')).toBe(MARKDOWN_CHUNKER_VERSION);
+    expect(MARKDOWN_CHUNKER_VERSION).toBeGreaterThan(1);
+  });
+
+  test('explicit chunker_version is honored on INSERT', async () => {
+    await engine.putPage('dream/explicit-chunker-version', {
+      type: 'concept',
+      title: 'Explicit version',
+      compiled_truth: 'Caller supplied a version.',
+      timeline: '',
+      chunker_version: 2,
+    });
+    expect(await readChunkerVersion('dream/explicit-chunker-version')).toBe(2);
+  });
+
+  test('re-put without chunker_version does not lower an already-current version', async () => {
+    await engine.putPage('dream/preserve-version', {
+      type: 'concept',
+      title: 'Preserve me',
+      compiled_truth: 'first write',
+      timeline: '',
+      chunker_version: MARKDOWN_CHUNKER_VERSION,
+    });
+    // Second write omits chunker_version; the UPDATE branch must not drop it
+    // below the current chunker version.
+    await engine.putPage('dream/preserve-version', {
+      type: 'concept',
+      title: 'Preserve me',
+      compiled_truth: 'second write',
+      timeline: '',
+    });
+    expect(await readChunkerVersion('dream/preserve-version')).toBe(MARKDOWN_CHUNKER_VERSION);
+  });
+});