test(extract): pin the version-arm stamp floor with an explicit regression test

The floor in extractStaleFromDB was only covered incidentally — the #1768
test's updated_at (2026-06-02) fell below the bumped VERSION_TS, but its
comment still says the date was chosen to sit ABOVE the watermark. A future
date 'repair' there would silently drop floor coverage. This test pins it
directly: pre-watermark page clears after --stale and stays cleared.
Verified fail-without-fix against master's extract.ts.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/cli.ts

@@ -24,7 +24,6 @@ import type { GBrainConfig } from './core/config.ts';
 import type { AIGatewayConfig } from './core/ai/types.ts';
 import type { BrainEngine } from './core/engine.ts';
 import { operations, OperationError } from './core/operations.ts';
-import { resolveSourceIdEngineFree } from './core/source-resolver.ts';
 import { formatVolunteeredPage } from './core/context/volunteer.ts';
 import type { Operation, OperationContext } from './core/operations.ts';
 import { shouldForceExitAfterMain, finishCliTeardown, flushThenExit, currentExitCode, setCliExitVerdict } from './core/cli-force-exit.ts';
@@ -383,15 +382,6 @@ async function main() {
     if (op.localOnly) {
       refuseThinClient(command, cfgPre!.remote_mcp!.mcp_url);
     }
-    // #2098: the local path resolves --source / GBRAIN_SOURCE / .gbrain-source
-    // inside makeContext (ctx.sourceId), which this route never reaches — so
-    // scope must be mapped onto the op's source_id wire param before the call.
-    try {
-      applyThinClientSourceScope(op, params);
-    } catch (e: unknown) {
-      console.error(e instanceof Error ? e.message : String(e));
-      process.exit(1);
-    }
     await runThinClientRouted(op, params, cfgPre!, cliOpts);
     return;
   }
@@ -812,60 +802,6 @@ export function parseOpArgs(op: Operation, args: string[]): Record<string, unkno
   return params;
 }
 
-/**
- * #2098: thin-client source scoping. Locally, --source / GBRAIN_SOURCE /
- * .gbrain-source resolve to ctx.sourceId in makeContext; the thin-client
- * route short-circuits before that, so `gbrain query --source X` against a
- * remote brain silently searched unscoped. This runs the engine-free tiers
- * (flag → env → dotfile; the DB-backed tiers can't run without an engine —
- * the server's grant scoping covers the rest) and maps the result onto the
- * op's `source_id` wire param.
- *
- * Ops that declare their OWN `source` param (facts add, etc.) are left
- * untouched — their --source is an op param, not scope. An explicit --source
- * on an op with no source_id wire param throws (loud beats silent drop);
- * ambient env/dotfile scope with nowhere to send it is ignored, matching the
- * pre-fix behavior for non-scopeable ops. Exported for tests.
- */
-// Ops whose `source_id` wire param is NOT read-scope semantics: get_skill's
-// source_id flips the lookup from host catalog to brain-resident-pack
-// (getResidentSkillDetail). Ambient env/dotfile scope must never leak into
-// these; an explicit --source-id still passes through untouched above.
-const NON_SCOPE_SOURCE_ID_OPS = new Set(['get_skill']);
-
-export function applyThinClientSourceScope(
-  op: Operation,
-  params: Record<string, unknown>,
-  cwd?: string,
-): void {
-  if ('source' in op.params) return; // the op owns --source; not a scope flag
-  const explicit = typeof params.source === 'string' && params.source.length > 0
-    ? (params.source as string)
-    : null;
-  delete params.source; // never a wire param on these ops — don't leak it
-  // Explicit per-call scope already on the wire wins over ambient tiers.
-  if (params.source_id !== undefined || params.all_sources === true) {
-    if (explicit) {
-      throw new Error('Pass either --source or --source-id/--all-sources, not both.');
-    }
-    return;
-  }
-  const resolved = resolveSourceIdEngineFree(explicit, cwd);
-  if (!resolved) return;
-  if (!('source_id' in op.params) || NON_SCOPE_SOURCE_ID_OPS.has(op.name)) {
-    if (explicit) {
-      const hint = NON_SCOPE_SOURCE_ID_OPS.has(op.name)
-        ? `(its source_id parameter is not a scope filter; pass --source-id explicitly if you mean it)`
-        : `(the remote op has no source_id parameter; the server scopes it to your grant)`;
-      throw new Error(
-        `gbrain ${op.cliHints?.name || op.name} does not accept --source on a thin-client install ${hint}.`,
-      );
-    }
-    return; // ambient env/dotfile scope with nowhere to send it
-  }
-  params.source_id = resolved;
-}
-
 async function makeContext(engine: BrainEngine, params: Record<string, unknown>): Promise<OperationContext> {
   // v0.31.8 (D11): resolve sourceId via the canonical 6-tier chain. Honors
   // --source / GBRAIN_SOURCE / .gbrain-source / path-match / brain default /

src/commands/extract.ts

@@ -1743,7 +1743,15 @@ async function extractStaleFromDB(
       // `page.updated_at.toISOString()` — the JS Date is ms-truncated, so the
       // µs-precision DB updated_at stayed strictly greater and the page never
       // cleared on Postgres. Stamping the exact value makes them equal.
-      processedRefs.push({ slug: page.slug, source_id: page.source_id, extractedAt: page.updated_at_iso });
+      //
+      // Version-arm floor: a page last edited BEFORE LINK_EXTRACTOR_VERSION_TS
+      // would otherwise be stamped below the version watermark and stay
+      // permanently stale (`links_extracted_at < versionTs` re-fires every run).
+      // Stamp max(updated_at, versionTs) — versionTs is always a past release
+      // date, so a concurrent edit's now() still exceeds the stamp and D4 holds.
+      // Tie at ms precision picks updated_at_iso (its µs ≥ versionTs's .000000).
+      const stampTs = new Date(page.updated_at_iso) >= new Date(versionTs) ? page.updated_at_iso : versionTs;
+      processedRefs.push({ slug: page.slug, source_id: page.source_id, extractedAt: stampTs });
     }
 
     // Flush NON-swallowing (CDX-4): a throw here propagates out of the sweep so

src/core/link-extraction.ts

@@ -28,7 +28,7 @@ import { ensureWellFormed } from './text-safe.ts';
  * OR updated_at > links_extracted_at`. It is an ISO-8601 string (NOT a number) —
  * the column is TIMESTAMPTZ and the predicate binds it as `::timestamptz`.
  */
-export const LINK_EXTRACTOR_VERSION_TS = '2026-05-31T00:00:00Z';
+export const LINK_EXTRACTOR_VERSION_TS = '2026-07-21T00:00:00Z';
 
 // ─── Entity references ──────────────────────────────────────────
 
@@ -80,10 +80,10 @@ export type LinkResolutionType = 'qualified' | 'unqualified';
  * Directory prefix whitelist. These are the top-level slug dirs the extractor
  * recognizes as entity references. Upstream canonical + our extensions:
  *   - Gbrain canonical: people, companies, meetings, concepts, deal, civic, project, source, media, yc, projects
- *   - Our domain extensions: tech, finance, personal, openclaw (domain-organized wikis)
+ *   - Our domain extensions: tech, finance, personal, openclaw, ops (domain-organized wikis)
  *   - Our entity prefix: entities (we kept some legacy entities/projects/ pages)
  */
-const DIR_PATTERN = '(?:people|companies|meetings|concepts|deal|civic|project|projects|source|media|yc|tech|finance|personal|openclaw|entities)';
+const DIR_PATTERN = '(?:people|companies|meetings|concepts|deal|civic|project|projects|source|media|yc|tech|finance|personal|openclaw|entities|ops)';
 
 /**
  * Match `[Name](path)` markdown links pointing to entity directories.
@@ -865,7 +865,16 @@ export function queryBasenameIndex(idx: Map<string, string[]>, name: string): st
   if (!name || typeof name !== 'string') return [];
   const trimmed = name.trim();
   if (!trimmed) return [];
-  const hit = idx.get(trimmed) ?? idx.get(trimmed.toLowerCase()) ?? idx.get(normalizeBasename(trimmed));
+  let hit = idx.get(trimmed) ?? idx.get(trimmed.toLowerCase()) ?? idx.get(normalizeBasename(trimmed));
+  // Issue #2576 bug 2: path-style refs (`runbooks/2026-05-01-x`) from dirs
+  // outside DIR_PATTERN reach here, but normalizeBasename strips slashes
+  // into a garbage key (`runbooks2026-05-01-x`) that can never hit the
+  // tail-keyed index. Fall back to the path tail so qualified refs resolve
+  // by basename like everything else.
+  if (!hit && trimmed.includes('/')) {
+    const tail = trimmed.slice(trimmed.lastIndexOf('/') + 1).trim();
+    if (tail) hit = idx.get(tail) ?? idx.get(tail.toLowerCase()) ?? idx.get(normalizeBasename(tail));
+  }
   return hit ? [...hit].sort(basenameSort) : [];
 }
 

src/core/source-resolver.ts

@@ -160,33 +160,6 @@ export async function resolveSourceId(
   return 'default';
 }
 
-/**
- * Engine-free tiers (1-3) of the resolution chain: explicit flag →
- * GBRAIN_SOURCE env → .gbrain-source dotfile walk. Used by the thin-client
- * CLI path (#2098), which has no local engine to run tiers 4-6 or
- * assertSourceExists against — the remote server enforces existence + grant.
- * Returns null when no engine-free tier fires.
- */
-export function resolveSourceIdEngineFree(
-  explicit: string | null | undefined,
-  cwd: string = process.cwd(),
-): string | null {
-  if (explicit) {
-    if (!SOURCE_ID_RE.test(explicit)) {
-      throw new Error(`Invalid --source value "${explicit}". Must match [a-z0-9-]{1,32}.`);
-    }
-    return explicit;
-  }
-  const env = process.env.GBRAIN_SOURCE;
-  if (env && env.length > 0) {
-    if (!SOURCE_ID_RE.test(env)) {
-      throw new Error(`Invalid GBRAIN_SOURCE value "${env}". Must match [a-z0-9-]{1,32}.`);
-    }
-    return env;
-  }
-  return readDotfileWalk(cwd);
-}
-
 /**
  * Returns the id of the SINGLE registered non-default source with a
  * local_path, when exactly one such row exists. Returns null when:

test/extract-stale.test.ts

@@ -209,6 +209,24 @@ describe('gbrain extract --stale', () => {
     expect(usRows[0]?.eq).toBe(true);
   });
 
+  test('version-arm floor: page edited BEFORE LINK_EXTRACTOR_VERSION_TS clears after --stale (issue #2576 bug 3)', async () => {
+    // A page whose updated_at predates the version watermark used to be
+    // stamped at its updated_at (< versionTs), so the version arm re-fired
+    // every run — permanently stale. The sweep now floors the stamp at
+    // versionTs. (The #1768 test above also covers this since the v0.42.x
+    // VERSION_TS bump moved its date below the watermark, but this pins the
+    // behavior explicitly so a date "repair" there can't drop coverage.)
+    await engine.putPage('people/alice', personPage('Alice'));
+    await engine.executeRaw(`UPDATE pages SET updated_at = '2000-01-01T00:00:00Z' WHERE slug = 'people/alice'`);
+    expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(1);
+
+    await runExtract(engine, ['--stale']);
+    // Pre-floor this stayed 1 forever (stamp < versionTs → version arm re-fires).
+    expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(0);
+    await runExtract(engine, ['--stale']);
+    expect(await engine.countStalePagesForExtraction({ versionTs: LINK_EXTRACTOR_VERSION_TS })).toBe(0);
+  });
+
   test('CDX-4 (D2): a link-flush throw aborts the sweep and leaves pages UNSTAMPED', async () => {
     await engine.putPage('people/alice', personPage('Alice'));
     await engine.putPage('companies/acme', companyPage('Acme', '[Alice](people/alice) founded [Acme](companies/acme).'));

test/link-extraction.test.ts

@@ -140,6 +140,15 @@ describe('extractEntityRefs', () => {
     expect(wikiRefs[0].needsResolution).toBe(true);
   });
 
+  test('recognizes ops/ qualified wikilinks (issue #2576 bug 2)', () => {
+    // `ops` was missing from DIR_PATTERN, so [[ops/...]] fell through to
+    // the generic 2c pass (needsResolution) instead of being a real ref.
+    const refs = extractEntityRefs('Deployed via [[ops/services/pointer-agent]].');
+    expect(refs.length).toBe(1);
+    expect(refs[0].slug).toBe('ops/services/pointer-agent');
+    expect(refs[0].needsResolution).toBeUndefined();
+  });
+
   test('skips qualified-syntax tokens (those belong to 2a)', () => {
     // [[wiki:topics/ai]] looks like 2a's qualified shape — even though
     // it wouldn't satisfy DIR_PATTERN, 2c must not claim it either
@@ -1069,6 +1078,15 @@ describe('makeResolver — fallback chain', () => {
     ]);
   });
 
+  test('resolveBasenameMatches: path-style ref falls back to the tail (issue #2576 bug 2)', async () => {
+    // normalizeBasename strips slashes, so `runbooks/2026-05-01-pointer-agent`
+    // used to normalize to a garbage key that never hit the tail-keyed index.
+    const engine = makeFakeEngineWithSlugs(['ops/changes/2026-05-01-pointer-agent']);
+    const r = makeResolver(engine);
+    expect(await r.resolveBasenameMatches!('runbooks/2026-05-01-pointer-agent'))
+      .toEqual(['ops/changes/2026-05-01-pointer-agent']);
+  });
+
   test('resolveBasenameMatches: case-insensitive fallback', async () => {
     const engine = makeFakeEngineWithSlugs(['companies/fast-weigh']);
     const r = makeResolver(engine);

test/thin-client-source-scope.test.ts

@@ -1,142 +0,0 @@
-/**
- * #2098: thin-client routing dropped --source / GBRAIN_SOURCE / .gbrain-source.
- *
- * The local CLI path resolves source scope in makeContext (ctx.sourceId); the
- * thin-client route short-circuits before that and sent params verbatim, so
- * `gbrain query --source X` against a remote brain silently searched unscoped
- * (the server op ignores the unknown `source` key).
- *
- * applyThinClientSourceScope runs the engine-free tiers (flag → env → dotfile)
- * and maps the result onto the op's `source_id` wire param. These tests fail
- * without the fix (params.source_id stays undefined / params.source leaks).
- */
-
-import { describe, test, expect } from 'bun:test';
-import { mkdtempSync, rmSync, writeFileSync } from 'fs';
-import { join } from 'path';
-import { tmpdir } from 'os';
-import { applyThinClientSourceScope, parseOpArgs } from '../src/cli.ts';
-import { operationsByName } from '../src/core/operations.ts';
-import { withEnv } from './helpers/with-env.ts';
-
-const queryOp = operationsByName.query;
-
-describe('applyThinClientSourceScope (#2098)', () => {
-  test('--source maps onto the query op wire param source_id', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['find things', '--source', 'wiki']);
-      expect(params.source).toBe('wiki'); // pre-fix state: wrong key
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('wiki');
-      expect('source' in params).toBe(false); // never leaks the unknown key
-    });
-  });
-
-  test('GBRAIN_SOURCE env tier fires when no flag is passed', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const params = parseOpArgs(queryOp, ['find things']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('gstack');
-    });
-  });
-
-  test('.gbrain-source dotfile tier fires when flag and env are absent', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const tmp = mkdtempSync(join(tmpdir(), 'gbrain-thin-scope-'));
-      try {
-        writeFileSync(join(tmp, '.gbrain-source'), 'essays\n');
-        const params = parseOpArgs(queryOp, ['find things']);
-        applyThinClientSourceScope(queryOp, params, tmp);
-        expect(params.source_id).toBe('essays');
-      } finally {
-        rmSync(tmp, { recursive: true, force: true });
-      }
-    });
-  });
-
-  test('explicit --source-id on the wire wins over ambient env scope', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const params = parseOpArgs(queryOp, ['find things', '--source-id', 'wiki']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBe('wiki');
-    });
-  });
-
-  test('--source together with --source-id is rejected loudly', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['q', '--source', 'a', '--source-id', 'b']);
-      expect(() => applyThinClientSourceScope(queryOp, params, '/')).toThrow(/not both/);
-    });
-  });
-
-  test('invalid --source value is rejected loudly', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['q', '--source', 'Bad_Value!']);
-      expect(() => applyThinClientSourceScope(queryOp, params, '/')).toThrow(/Invalid --source/);
-    });
-  });
-
-  test('--source on an op with no source_id wire param errors instead of silently dropping', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.add_tag;
-      expect('source_id' in op.params).toBe(false);
-      const params = { slug: 'x', tag: 'y', source: 'wiki' };
-      expect(() => applyThinClientSourceScope(op, params, '/')).toThrow(/--source/);
-    });
-  });
-
-  test('ambient env scope on an op with no source_id wire param is ignored (no throw)', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'wiki' }, () => {
-      const op = operationsByName.add_tag;
-      const params: Record<string, unknown> = { slug: 'x', tag: 'y' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-
-  test('ops that declare their OWN source param are left untouched', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.put_raw_data;
-      expect('source' in op.params).toBe(true);
-      const params: Record<string, unknown> = { slug: 'x', source: 'crustdata', data: {} };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source).toBe('crustdata');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-
-  test('get_skill: ambient scope never leaks into its non-scope source_id param', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'wiki' }, () => {
-      const op = operationsByName.get_skill;
-      expect('source_id' in op.params).toBe(true); // has the param, but it is a mode switch
-      const params: Record<string, unknown> = { name: 'ingest' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBeUndefined(); // would flip host catalog → brain-pack lookup
-    });
-  });
-
-  test('get_skill: explicit --source errors instead of masquerading as --source-id', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const op = operationsByName.get_skill;
-      const params: Record<string, unknown> = { name: 'ingest', source: 'wiki' };
-      expect(() => applyThinClientSourceScope(op, params, '/')).toThrow(/--source-id/);
-    });
-  });
-
-  test('get_skill: explicit --source-id passes through untouched', async () => {
-    await withEnv({ GBRAIN_SOURCE: 'gstack' }, () => {
-      const op = operationsByName.get_skill;
-      const params: Record<string, unknown> = { name: 'ingest', source_id: 'wiki' };
-      applyThinClientSourceScope(op, params, '/');
-      expect(params.source_id).toBe('wiki');
-    });
-  });
-
-  test('no scope from any tier leaves params unchanged', async () => {
-    await withEnv({ GBRAIN_SOURCE: undefined }, () => {
-      const params = parseOpArgs(queryOp, ['find things']);
-      applyThinClientSourceScope(queryOp, params, '/');
-      expect(params.source_id).toBeUndefined();
-    });
-  });
-});