test: reset gateway in afterAll in cli-multimodal-integration — cross-file config leak

Review hardening on #3116: the file resets the gateway in beforeEach but
left it configured after the last test, leaking a multimodal gateway
config into whatever file runs next in its shard (same class as the
adaptive-embed-batch leak fixed on master).

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

docs/architecture/KEY_FILES.md

@@ -79,7 +79,7 @@ per-release `**vX.Y.Z:**` narration — CI enforces this
 - `docs/architecture/RETRIEVAL.md` + `docs/architecture/RETRIEVAL_MAXPOOL_INCIDENT.md` — retrieval-pipeline architecture reference + the named-thing-miss incident write-up (root cause, the five-layer fix, the eval that pins it).
 - `src/core/types.ts` extension + `src/core/operations.ts:search` + `src/core/import-file.ts` + `src/cli.ts` + `src/core/search/telemetry.ts` — the wiring layer for the retrieval cathedral. `SearchResult` gains `evidence`, `create_safety`, `title_match_boost`, `alias_hit` (all optional; evidence/create_safety reference the union types in `evidence.ts`). The `search` MCP op uses a cheap-hybrid path by default and accepts a per-call `mode` (conservative|balanced|tokenmax) honored ONLY for trusted/local callers (`resolvePerCallMode(ctx, ...)` — remote callers use the configured mode so a remote provider can't force tokenmax spend); every search path stamps evidence fail-soft. `importFromContent` projects frontmatter `aliases:` into `page_aliases` via `normalizeAliasList` + `engine.setPageAliases` so new + changed pages register aliases at ingest. `src/cli.ts` adds the `gbrain search diagnose` dispatch (lazy import) and reconciles the `search` CLI path with the cheap-hybrid op. `src/core/search/telemetry.ts` extends the rollup with the rank-1 base_score drift signal (sum/count + 3 coarse buckets, aggregate not per-query), surfaced via `gbrain search stats`, backed by migration v111's `search_telemetry` columns. Tests: `test/cli-search-dispatch.test.ts`, `test/search/per-call-mode.test.ts`, `test/search/telemetry-rank1.test.ts`, `test/search/title-boost-stage.test.ts`, `test/search/alias-hop.test.ts`, `test/search/evidence.test.ts`, `test/search/searchvector-maxpool.test.ts`, `test/search/pre-migration-failopen.test.ts`.
 - `src/commands/eval.ts` — `gbrain eval` command: single-run table + A/B config comparison. Sub-subcommand dispatch on `args[0]` routes `gbrain eval export` + `gbrain eval prune` + `gbrain eval replay` into session-capture handlers; bare `gbrain eval --qrels …` fall-through preserves the legacy IR-metrics flow. `gbrain eval cross-modal` is in the dispatch (the user-facing path is the cli.ts no-DB branch — `src/commands/eval.ts:cross-modal` only fires when callers re-enter with an existing engine).
-- `src/commands/eval-cross-modal.ts` — multi-model quality gate. Three different-provider frontier models score the OUTPUT against the TASK on a 5-dim list. Verdict `pass` (exit 0) / `fail` (exit 1) / `inconclusive` (exit 2; <2/3 model successes). Reuses `src/core/ai/gateway.ts:chat()` so config/auth/aliasing comes from the gateway recipe registry — no parallel provider stack. Self-configures the gateway (`configureGateway(loadConfig() + process.env)`) since the cli.ts dispatch bypasses `connectEngine()`. Default cycles 3 in TTY, 1 in non-TTY (partial cost guardrail) via the shared `resolveCycleDefault(explicit, isTty)` in `src/core/eval/cycle-default.ts`; the cost-estimate banner appends `cycleDefaultSuffix(...)` (`for 1 cycle(s) (non-interactive default; --cycles N for more)`) when the value is the silent non-TTY fallback, so the 1-vs-3 difference isn't hidden. Receipts land at `gbrainPath('eval-receipts')/<slug>-<sha8-of-output>.json`. `--batch <jsonl> [--limit N] [--concurrent N] [--max-usd FLOAT] [--yes]` fans out cross-modal scoring across a LongMemEval-shape JSONL; mutually exclusive with `--task` (fail-fast usage error if both set); filters `kind: "by_type_summary"` rows; pre-flight cost estimate refuses if `> --max-usd` without `--yes` (default cap 5.00 USD). Semaphore-bounded fan-out via inline `runWithLimit<T>(items, limit, fn)` (exported for unit tests): max N questions in-flight × 3 model slots = ceiling of 3N parallel API calls (default `--concurrent 3` → 9). Per-question receipts land in a per-batch tempdir and are deleted at end of run; the summary receipt inlines per-question verdicts as JSON, not file paths. Exit precedence (batch-level policy, NOT inherited from aggregate.ts): ERROR > FAIL > INCONCLUSIVE > PASS. DI seam: `runEvalCrossModal(args, opts?: {runEval?: typeof runEval})` mirrors `runEvalLongMemEval(args, {client?})`; tests pass `opts.runEval` to bypass real LLM calls AND the gateway availability check. Pinned by `test/eval-cross-modal-batch.test.ts`.
+- `src/commands/eval-cross-modal.ts` — multi-model quality gate. Three different-provider frontier models score the OUTPUT against the TASK on a 5-dim list. Verdict `pass` (exit 0) / `fail` (exit 1) / `inconclusive` (exit 2; <2/3 model successes). Reuses `src/core/ai/gateway.ts:chat()` so config/auth/aliasing comes from the gateway recipe registry — no parallel provider stack. Self-configures the gateway via `buildGatewayConfig(loadConfig() ?? {})` since the cli.ts dispatch bypasses `connectEngine()`, so file-plane keys, env base URLs, and provider_chat_options follow the same adapter path as runtime. Default cycles 3 in TTY, 1 in non-TTY (partial cost guardrail) via the shared `resolveCycleDefault(explicit, isTty)` in `src/core/eval/cycle-default.ts`; the cost-estimate banner appends `cycleDefaultSuffix(...)` (`for 1 cycle(s) (non-interactive default; --cycles N for more)`) when the value is the silent non-TTY fallback, so the 1-vs-3 difference isn't hidden. Receipts land at `gbrainPath('eval-receipts')/<slug>-<sha8-of-output>.json`. `--batch <jsonl> [--limit N] [--concurrent N] [--max-usd FLOAT] [--yes]` fans out cross-modal scoring across a LongMemEval-shape JSONL; mutually exclusive with `--task` (fail-fast usage error if both set); filters `kind: "by_type_summary"` rows; pre-flight cost estimate refuses if `> --max-usd` without `--yes` (default cap 5.00 USD). Semaphore-bounded fan-out via inline `runWithLimit<T>(items, limit, fn)` (exported for unit tests): max N questions in-flight × 3 model slots = ceiling of 3N parallel API calls (default `--concurrent 3` → 9). Per-question receipts land in a per-batch tempdir and are deleted at end of run; the summary receipt inlines per-question verdicts as JSON, not file paths. Exit precedence (batch-level policy, NOT inherited from aggregate.ts): ERROR > FAIL > INCONCLUSIVE > PASS. DI seam: `runEvalCrossModal(args, opts?: {runEval?: typeof runEval})` mirrors `runEvalLongMemEval(args, {client?})`; tests pass `opts.runEval` to bypass real LLM calls AND the gateway availability check. Pinned by `test/eval-cross-modal-batch.test.ts`.
 - `src/core/eval/cycle-default.ts` — single source of truth for the eval cycle-count default. Exports `DEFAULT_CYCLES_TTY = 3`, `DEFAULT_CYCLES_NONTTY = 1`, `resolveCycleDefault(explicit, isTty): {cycles, usedNonTtyDefault}`, and `cycleDefaultSuffix(r)` (returns ` (non-interactive default; --cycles N for more)` only when the non-TTY default was applied, else `''`). Consumed by `eval-cross-modal.ts`, `eval-takes-quality.ts` (run + regress), and `takes-quality-eval/runner.ts` (core uses only the constant — library stays TTY-agnostic; the CLI owns the TTY=3 upgrade + banner annotation). `eval-suspected-contradictions.ts` applies the same transparency to its `$5`/`$1` budget default via a `budgetUsdExplicit` flag (the budget is overwritten in-place so explicitness can't be inferred post-hoc). Not shared with `resolveWorkersWithClamp` (different domain, no engine, no dedup). Pinned by `test/eval/cycle-default.test.ts`, `test/eval-suspected-contradictions-budget-default.test.ts`.
 - `src/core/cross-modal-eval/json-repair.ts` — `parseModelJSON(raw)` named export with a 4-strategy fallback chain (direct parse → fence-strip → trailing-comma + single-quote + embedded-newline repair → regex nuclear option). Adversarial input throws rather than fabricating scores — the aggregator treats a throw as "this model contributed nothing this cycle" so the gate stays correct at >=2/3 successes.
 - `src/core/cross-modal-eval/aggregate.ts` — pure verdict logic. Pass criterion: `(successes >= 2) AND (every dim mean >= 7) AND (every dim min across models >= 5)`. Inconclusive when <2/3 models returned parseable scores (regression guard for the v1 `Object.values({}).every(...) === true` empty-array PASS bug).
@@ -138,7 +138,7 @@ per-release `**vX.Y.Z:**` narration — CI enforces this
 - `src/core/minions/queue.ts` extension — `MinionQueue.add()` rejects `subagent` jobs whose `data.model` resolves via `isAnthropicProvider()` to a non-Anthropic provider. Lazy-imports `model-config.ts` to avoid pulling engine types into queue's eager-load surface. Layer 1 of the three-layer subagent provider enforcement (layers 2+3: `model-config.ts:enforceSubagentAnthropic` runtime fallback + `src/commands/doctor.ts` `subagent_provider` check). Pinned by `test/agent-cli.test.ts`.
 - `src/commands/models.ts` — `gbrain models [--json]` read-only routing dashboard: prints tier defaults (`utility`/`reasoning`/`deep`/`subagent`), the resolved value for each (re-walking the resolution chain), every per-task override (11 `PER_TASK_KEYS`: `models.dream.synthesize`, `models.dream.patterns`, `models.drift`, `models.auto_think`, `models.think`, `models.subagent`, `facts.extraction_model`, `models.eval.longmemeval`, `models.expansion`, `models.chat`, `models.dream.synthesize_verdict`), the alias map, and a source-of-truth column (`default` / `config: <key>` / `env: <VAR>`). `gbrain models doctor [--skip=<provider>] [--json]` fires a 1-token `gateway.chat()` probe against each configured chat + expansion model and classifies failures into `{model_not_found, auth, rate_limit, network, unknown}`. Wired into `cli.ts` dispatch + `CLI_ONLY` set. A zero-token `embedding_config` probe runs FIRST, before any chat/expansion probes spend money: `probeEmbeddingConfig()` reads `getEmbeddingModel()` + `getEmbeddingDimensions()` and (for Voyage flexible-dim models) checks `isValidVoyageOutputDim(dims)` against `VOYAGE_VALID_OUTPUT_DIMS`. `ProbeStatus` variant `'config'` + optional `fix?: string` on `ProbeResult` surface a paste-ready `gbrain config set ...` line in human + JSON output; touchpoint label `'embedding_config'` joins `'chat'` and `'expansion'`.
 - `src/core/init-embed-check.ts` — embedding-key validation at `gbrain init`. `runInitEmbedCheck(opts)` runs a config-only `diagnoseEmbedding` (catches a missing key for ANY provider) plus a best-effort `liveTestEmbed` (1-token `gateway.embed(['probe'], {inputType:'query', abortSignal})`, 5s `AbortController` timeout, never throws — catches an invalid/expired key). Loud warning to stderr; init still exits 0 (`--no-embedding` is the deferred-setup escape; `--skip-embed-check` / `GBRAIN_INIT_SKIP_EMBED_CHECK=1` skip the check). Builds the effective env (`process.env` + file-plane `openai/anthropic/zeroentropy_api_key` from `loadConfigFileOnly()` + `opts.apiKey`) and configures the gateway via `buildGatewayConfig` before diagnose/probe, so the check sees the same keys AND provider base URLs runtime will (no false "missing key" for config.json-keyed users; the probe hits the right endpoint). Init-specific warning text names `--no-embedding` / `--skip-embed-check`, not the sync-flavored `--no-embed`. Wired into `initPGLite` + `initPostgres` in `src/commands/init.ts`, with the result added to the `--json` envelope as `embedding_check {ok, reason?, live_ok?}`. Pinned by `test/init-embed-check.test.ts` (hermetic via the gateway embed-transport seam + `withEnv`).
-- `src/core/ai/build-gateway-config.ts` — `buildGatewayConfig(c: GBrainConfig): AIGatewayConfig`, extracted from `src/cli.ts` (which re-exports it for back-compat). Lets core modules (`init-embed-check.ts`) reuse it without importing the CLI entrypoint. Single owner of folding file-plane API keys (openai/anthropic/zeroentropy) into the gateway env and threading local-server `*_BASE_URL` env vars into base_urls. `process.env` wins EXCEPT empty-string / undefined values are dropped before the merge, so an injected empty `ANTHROPIC_API_KEY=''` (Claude Code neuters subprocess LLM calls this way) can't clobber a valid config-plane key; `'0'` / `'false'` are preserved. Pinned by `test/ai/build-gateway-config.test.ts`.
+- `src/core/ai/build-gateway-config.ts` — `buildGatewayConfig(c: GBrainConfig): AIGatewayConfig`, extracted from `src/cli.ts` (which re-exports it for back-compat). Single owner of translating stored config into gateway config — consumed by CLI runtime, init (`gbrain init`'s three configureGateway sites), `init-embed-check.ts`, the eval commands (cross-modal, takes-quality), provider diagnostics, and the in-process migration path. Folds file-plane API keys (openai/anthropic/zeroentropy/openrouter) into the gateway env and threads local-server `*_BASE_URL` env vars into base_urls; caller-provided `provider_base_urls` config wins over env base URLs. `process.env` wins EXCEPT empty-string / undefined values are dropped before the merge, so an injected empty `ANTHROPIC_API_KEY=''` (Claude Code neuters subprocess LLM calls this way) can't clobber a valid config-plane key; `'0'` / `'false'` are preserved. Pinned by `test/ai/build-gateway-config.test.ts`.
 - `src/commands/doctor.ts` extension — `subagent_provider` check (layer 3 of 3). Warns when `models.tier.subagent` is explicitly set non-Anthropic (message names the bad value + paste-ready fix `gbrain config set models.tier.subagent anthropic:claude-sonnet-4-6`); also warns when `models.default` would sneak `subagent` into a non-Anthropic provider via tier inheritance. OK when subagent tier resolves to Anthropic. Tests in `test/doctor.test.ts`.
 - `src/core/skill-trigger-index.ts` — Shared loader that unions per-skill SKILL.md frontmatter `triggers:` with curated RESOLVER.md / AGENTS.md rows from `skillsDir` AND the parent dir (preserves the OpenClaw workspace-root layout). UNION semantics: explicit RESOLVER.md rows ADD to frontmatter triggers (don't replace). Dedup keyed on `(skillPath, trigger.trim().toLowerCase())`. Three consumers fold through this primitive — `checkResolvable`, `runRoutingEvalCli`, `mounts-cache.composeResolvers` — so fixing frontmatter reaches all of them. Exports `loadSkillTriggerIndex(skillsDir): SkillTriggerEntry[]`, `entriesToResolverContent(entries): string` (synthesizes a markdown-table resolver string for `runRoutingEval`'s string-content API), `findPrimaryResolverPath(skillsDir): string | null`, the `FRONTMATTER_SECTION` constant, and `_resetWarnedSkillsForTests`. Skip rules: non-directory entries, `_*`/`.*` prefixes, `conventions/`+`migrations/` subdirs, skills with no `SKILL.md` (deprecated `install/` graceful-skipped), no `triggers:` array, or malformed YAML (warn-once + skip). Reuses `parseSkillFrontmatter` from `src/core/skill-frontmatter.ts` (regex-based, not full YAML). Pinned by `test/skill-trigger-index.test.ts` (18 hermetic cases). CI gate `bun run check:resolver` (= `bun src/cli.ts check-resolvable --strict --skills-dir skills/`) wired into `bun run verify`.
 - `src/core/skill-catalog.ts` — host-repo skill catalog backing the MCP `list_skills` / `get_skill` ops. Lets a thin MCP client (Codex desktop, Claude Code, Claude Cowork, Perplexity) DISCOVER + FOLLOW the agent repo's fat-markdown skills over `gbrain serve` — a skill is prose, so "using" one = fetching its body then calling the gbrain MCP tools the server already exposes. Read-scope, NOT localOnly (defensible only via the full mitigation stack): (1) **publish gate** — `assertPublishEnabled(ctx, publishSkills)`; remote callers require `mcp.publish_skills === true`, default-OFF so an upgrade never silently grants existing read tokens host-skill read; local callers (`ctx.remote === false`) always pass. (2) **path confinement** — `assertSkillNameShape` rejects separators/`..`/null/space before any FS access; the client `name` is a manifest LOOKUP KEY (via `loadOrDeriveManifest`), never a raw path segment; `confineManifestPath` does realpath + relative-containment + `SKILL.md`-regular-file check on EVERY entry (defeats poisoned manifest.json `path`, symlink/`..` escape). (3) **frontmatter allowlist** — `GetSkillResult.frontmatter` projects a safe subset; private `writes_to` + `sources` dropped. (4) **prose-only + 256KB cap** (`MAX_SKILL_MD_BYTES`, env `GBRAIN_MAX_SKILL_MD_BYTES`), size-checked twice (statSync + UTF-8 byte length). (5) **no install_path serve for remote** — remote callers use `autoDetectSkillsDir` (no install-path tier) so a hosted gbrain with no agent repo returns `storage_error`; local callers use `autoDetectSkillsDirReadOnly`. (6) MCP rate-limiter caps call rate. Config reads honor BOTH planes: `readMcpPublishSkills` / `readMcpSkillsDir` prefer the DB plane (`engine.getConfig`) over the file plane (`ctx.config.mcp`). Tool-honesty: `crossReferenceTools(declared, ctx)` splits a skill's declared `tools:` into `usable_tools` vs `unavailable_tools`; `buildSkillCatalog`'s `instructions` envelope (`SKILL_CATALOG_INSTRUCTIONS`) carries the "these are prose, follow-then-call-tools" protocol. Skills are host-filesystem repo-global — `sourceScopeOpts(ctx)` / `ctx.brainId` deliberately do NOT apply. `buildSkillCatalog` is resilient (one malformed/escaping skill is skipped, never throws). Config keys in `src/core/config.ts`: `GBrainConfig.mcp?: { publish_skills?, skills_dir? }` + `KNOWN_CONFIG_KEYS` entries `mcp.publish_skills`/`mcp.publish_skills_prompted`/`mcp.skills_dir` + `mcp.` prefix in `KNOWN_CONFIG_KEY_PREFIXES`. `src/commands/init.ts` writes `config.mcp = { publish_skills: true, ... }` for new installs (existing config wins on re-init). `src/commands/upgrade.ts:runPostUpgrade` adds a one-time consent prompt (gated by `mcp.publish_skills_prompted`; existing installs stay OFF until owner opts in). Two ops register in `src/core/operations.ts` (`list_skills` with optional `section` filter + `cliHints:{name:'skills'}`; `get_skill` taking `name` + `cliHints:{name:'skill', positional:['name']}`) and dynamically import this module to avoid the import cycle (skill-catalog statically imports the `operations` array). Descriptions in `src/core/operations-descriptions.ts` (`LIST_SKILLS_DESCRIPTION`, `GET_SKILL_DESCRIPTION`, `SKILL_CATALOG_INSTRUCTIONS`, `SKILL_CLIENT_GUIDANCE`), pinned by `test/operations-descriptions.test.ts`. CLI: `gbrain skills` / `gbrain skill <name>`. Pinned by `test/skill-catalog.test.ts`, `test/skill-catalog-security.test.ts` (path-confinement / poisoned-manifest / symlink-escape), `test/skill-catalog-transports.test.ts` (publish-gate + remote-vs-local) over `test/fixtures/skill-catalog/`.

src/commands/doctor.ts

@@ -820,10 +820,6 @@ export async function doctorReportRemote(engine: BrainEngine): Promise<DoctorRep
   // v0.42.x (#1794, 4A): pool-budget nudge when GBRAIN_MAX_CONNECTIONS is set.
   checks.push(await checkPoolBudget(engine));
 
-  // #2552: warn when an explicit embed-concurrency override fans out against
-  // a local single-slot embedding endpoint (silent backfill starvation).
-  checks.push(await checkEmbedConcurrency());
-
   // v0.42.7 (#1696): link-extraction lag. Strictly SQL (single indexed COUNT),
   // safe on the thin-client/remote path — remote operators on checkout-less
   // Postgres brains are exactly who can't otherwise see the extraction backlog.
@@ -3819,61 +3815,6 @@ export function computePoolBudgetCheck(
   };
 }
 
-/**
- * #2552: warn when an explicit GBRAIN_EMBED_CONCURRENCY override fans out
- * against a local single-slot embedding endpoint (Ollama / llama-server /
- * localhost base URL). Requests serialize on the one loaded model, so N
- * parallel pages multiply latency xN and can exceed the fetch timeout with
- * no surfaced error — the backfill silently starves. (When the env var is
- * unset, embed auto-caps at LOCAL_EMBED_CONCURRENCY_CAP and this check
- * reports ok.) Pure; exported for tests.
- */
-export function computeEmbedConcurrencyCheck(
-  isLocalEndpoint: boolean,
-  envValue: string | undefined,
-  localCap: number,
-): Check {
-  const name = 'embed_concurrency';
-  if (!isLocalEndpoint) {
-    return { name, status: 'ok', message: 'Embedding endpoint is not a local inference server — cloud concurrency defaults apply.' };
-  }
-  const parsed = envValue ? parseInt(envValue, 10) : NaN;
-  if (envValue && Number.isFinite(parsed) && parsed > localCap) {
-    return {
-      name,
-      status: 'warn',
-      message:
-        `GBRAIN_EMBED_CONCURRENCY=${parsed} against a local embedding endpoint. ` +
-        `Local inference servers serialize requests, so ${parsed} parallel pages multiply ` +
-        `latency x${parsed} and can exceed the fetch timeout — the embed backfill stalls ` +
-        `with no error. Unset GBRAIN_EMBED_CONCURRENCY (auto-caps at ${localCap}) or set it <= ${localCap}.`,
-    };
-  }
-  return {
-    name,
-    status: 'ok',
-    message: `Local embedding endpoint detected; embed concurrency capped at ${envValue ? parsed : localCap}.`,
-  };
-}
-
-/** Thin gateway/env wrapper over `computeEmbedConcurrencyCheck`. */
-export async function checkEmbedConcurrency(): Promise<Check> {
-  try {
-    const { isLocalEmbeddingEndpoint, LOCAL_EMBED_CONCURRENCY_CAP } = await import('../core/ai/gateway.ts');
-    return computeEmbedConcurrencyCheck(
-      isLocalEmbeddingEndpoint(),
-      process.env.GBRAIN_EMBED_CONCURRENCY,
-      LOCAL_EMBED_CONCURRENCY_CAP,
-    );
-  } catch (err) {
-    return {
-      name: 'embed_concurrency',
-      status: 'ok',
-      message: `Skipped (${err instanceof Error ? err.message : String(err)})`,
-    };
-  }
-}
-
 /** Thin env/engine wrapper over `computePoolBudgetCheck`. */
 export async function checkPoolBudget(_engine: BrainEngine): Promise<Check> {
   try {

src/commands/embed.ts

@@ -1,6 +1,5 @@
 import type { BrainEngine } from '../core/engine.ts';
 import { embedBatch, currentEmbeddingSignature } from '../core/embedding.ts';
-import { isLocalEmbeddingEndpoint, LOCAL_EMBED_CONCURRENCY_CAP } from '../core/ai/gateway.ts';
 import type { ChunkInput } from '../core/types.ts';
 import { chunkText } from '../core/chunkers/recursive.ts';
 import { createProgress, type ProgressReporter } from '../core/progress.ts';
@@ -177,31 +176,6 @@ export class EmbeddingDimMismatchError extends Error {
   }
 }
 
-/**
- * #2552: resolve the bulk-embed worker count. Env override or the
- * cloud-tuned default of 20 — but when the operator did NOT set
- * GBRAIN_EMBED_CONCURRENCY and the embedding endpoint is a local inference
- * server (Ollama / llama-server / localhost base URL), cap at
- * LOCAL_EMBED_CONCURRENCY_CAP: 20 parallel pages against a single-slot
- * server serialize on the one loaded model, multiply latency x20 past the
- * fetch timeout, and starve the backfill with no surfaced error. An
- * explicit env value always wins (`gbrain doctor` warns instead).
- * Pacing only ever LOWERS concurrency (Codex P2).
- */
-export function resolveEmbedConcurrency(paceMaxConcurrency?: number): number {
-  const envSet = !!process.env.GBRAIN_EMBED_CONCURRENCY;
-  const base = parseInt(process.env.GBRAIN_EMBED_CONCURRENCY || '20', 10);
-  let resolved = base;
-  if (!envSet && isLocalEmbeddingEndpoint() && base > LOCAL_EMBED_CONCURRENCY_CAP) {
-    resolved = LOCAL_EMBED_CONCURRENCY_CAP;
-    serr(
-      `[embed] local embedding endpoint detected — capping concurrency at ` +
-      `${LOCAL_EMBED_CONCURRENCY_CAP} (set GBRAIN_EMBED_CONCURRENCY to override)`,
-    );
-  }
-  return paceMaxConcurrency ? Math.min(resolved, paceMaxConcurrency) : resolved;
-}
-
 /**
  * Pre-flight check: read the actual schema column dim and compare to the
  * gateway's resolved dim. Throws `EmbeddingDimMismatchError` on mismatch
@@ -703,8 +677,10 @@ async function embedAll(
   // Paced runs lower this to the resolved cap (the real lever vs pooler-slot
   // starvation); unpaced keeps the env/default 20. Codex P2: only ever LOWER —
   // never raise above an operator's existing env cap.
-  // #2552: local endpoints auto-cap — see resolveEmbedConcurrency.
-  const CONCURRENCY = resolveEmbedConcurrency(staleOpts?.paceMaxConcurrency);
+  const BASE_CONCURRENCY = parseInt(process.env.GBRAIN_EMBED_CONCURRENCY || '20', 10);
+  const CONCURRENCY = staleOpts?.paceMaxConcurrency
+    ? Math.min(BASE_CONCURRENCY, staleOpts.paceMaxConcurrency)
+    : BASE_CONCURRENCY;
 
   async function embedOnePage(page: typeof pages[number]) {
     // #1737: bail before doing any work for this page if the run was aborted.
@@ -879,8 +855,10 @@ async function embedAllStale(
   // Paced runs lower concurrency to the resolved cap (E-1: worker count IS the
   // lever on this single pool, no separate permit). Codex P2: pacing only ever
   // LOWERS concurrency — never raise above an operator's existing env cap.
-  // #2552: local endpoints auto-cap — see resolveEmbedConcurrency.
-  const CONCURRENCY = resolveEmbedConcurrency(staleOpts?.paceMaxConcurrency);
+  const BASE_CONCURRENCY = parseInt(process.env.GBRAIN_EMBED_CONCURRENCY || '20', 10);
+  const CONCURRENCY = staleOpts?.paceMaxConcurrency
+    ? Math.min(BASE_CONCURRENCY, staleOpts.paceMaxConcurrency)
+    : BASE_CONCURRENCY;
   const pacer = staleOpts?.pacer ?? createNoopPacer();
 
   // D3 + D3a + D8: wall-clock budget. 30 min default; env override.

src/commands/eval-cross-modal.ts

@@ -21,7 +21,8 @@ import { join } from 'path';
 import { tmpdir } from 'os';
 import { createHash } from 'crypto';
 
-import { gbrainPath, loadConfig } from '../core/config.ts';
+import { gbrainPath, loadConfig, type GBrainConfig } from '../core/config.ts';
+import { buildGatewayConfig } from '../core/ai/build-gateway-config.ts';
 import { configureGateway, isAvailable } from '../core/ai/gateway.ts';
 import { runWithLimit } from '../core/worker-pool.ts';
 import { resolveCycleDefault, cycleDefaultSuffix } from '../core/eval/cycle-default.ts';
@@ -264,32 +265,11 @@ function isTTY(): boolean {
  * Returns true on success; false (and prints a hint) when no config is found.
  */
 function configureGatewayForCli(): boolean {
+  // Route through buildGatewayConfig (the single adapter seam) so file-plane
+  // API keys, env base URLs, and provider_chat_options follow the same
+  // precedence as the runtime path. No config file is fine — env alone serves.
   const config = loadConfig();
-  if (!config) {
-    // No config file is fine for the eval command — env vars alone may serve.
-    // We still call configureGateway so gateway recipes can read the env map.
-    configureGateway({
-      embedding_model: undefined,
-      embedding_dimensions: undefined,
-      expansion_model: undefined,
-      chat_model: undefined,
-      chat_fallback_chain: undefined,
-      base_urls: undefined,
-      provider_chat_options: undefined,
-      env: { ...process.env },
-    });
-    return true;
-  }
-  configureGateway({
-    embedding_model: config.embedding_model,
-    embedding_dimensions: config.embedding_dimensions,
-    expansion_model: config.expansion_model,
-    chat_model: config.chat_model,
-    chat_fallback_chain: config.chat_fallback_chain,
-    base_urls: config.provider_base_urls,
-    provider_chat_options: config.provider_chat_options,
-    env: { ...process.env },
-  });
+  configureGateway(buildGatewayConfig(config ?? ({} as GBrainConfig)));
   return true;
 }
 

src/commands/eval-takes-quality.ts

@@ -21,7 +21,8 @@
  */
 import type { BrainEngine } from '../core/engine.ts';
 import { configureGateway } from '../core/ai/gateway.ts';
-import { loadConfig } from '../core/config.ts';
+import { buildGatewayConfig } from '../core/ai/build-gateway-config.ts';
+import { loadConfig, type GBrainConfig } from '../core/config.ts';
 import { runEval, DEFAULT_MODEL_PANEL } from '../core/takes-quality-eval/runner.ts';
 import { resolveCycleDefault, cycleDefaultSuffix } from '../core/eval/cycle-default.ts';
 import { writeReceipt } from '../core/takes-quality-eval/receipt-write.ts';
@@ -127,8 +128,12 @@ export async function runReplayNoBrain(argv: string[]): Promise<number> {
 export async function runEvalTakesQuality(engine: BrainEngine, args: string[]): Promise<void> {
   // Self-configure the AI gateway (mirrors eval-cross-modal pattern). The
   // gateway needs config.ai_gateway + env vars; configureGateway reads both.
+  // Route through buildGatewayConfig: the old `{ ...cfg, ...process.env }`
+  // spread never populated the gateway's `env` field (the gateway NEVER reads
+  // process.env at call time), so availability checks saw no keys at all and
+  // file-plane API keys / provider base URLs were dropped.
   const cfg = loadConfig();
-  configureGateway({ ...cfg, ...(process.env as Record<string, string>) } as any);
+  configureGateway(buildGatewayConfig(cfg ?? ({} as GBrainConfig)));
 
   const { subcmd, argv, json } = parseSubcmd(args);
 

src/commands/init.ts

@@ -7,6 +7,7 @@ import { homedir } from 'os';
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = dirname(__filename);
 import { saveConfig, loadConfig, loadConfigFileOnly, toEngineConfig, gbrainPath, configPath, isThinClient, effectiveEnvDatabaseUrl, type GBrainConfig } from '../core/config.ts';
+import { buildGatewayConfig } from '../core/ai/build-gateway-config.ts';
 import { createEngine } from '../core/engine-factory.ts';
 import { discoverOAuth, mintClientCredentialsToken, smokeTestMcp } from '../core/remote-mcp-probe.ts';
 import { runInitEmbedCheck } from '../core/init-embed-check.ts';
@@ -722,7 +723,8 @@ async function configureGatewayWithMergedPrecedence(
   // pollutes config.json.
   const envOverlay = loadConfig() ?? ({} as GBrainConfig);
 
-  const merged = {
+  const merged: GBrainConfig = {
+    ...envOverlay,
     embedding_model: aiOpts?.embedding_model ?? envOverlay.embedding_model ?? existingFile.embedding_model,
     embedding_dimensions: aiOpts?.embedding_dimensions ?? envOverlay.embedding_dimensions ?? existingFile.embedding_dimensions,
     expansion_model: aiOpts?.expansion_model ?? envOverlay.expansion_model ?? existingFile.expansion_model,
@@ -730,13 +732,9 @@ async function configureGatewayWithMergedPrecedence(
   };
 
   const { configureGateway, getEmbeddingModel, getEmbeddingDimensions, getExpansionModel, getChatModel } = await import('../core/ai/gateway.ts');
-  configureGateway({
-    embedding_model: merged.embedding_model,
-    embedding_dimensions: merged.embedding_dimensions,
-    expansion_model: merged.expansion_model,
-    chat_model: merged.chat_model,
-    env: { ...process.env },
-  });
+  // buildGatewayConfig (the single adapter seam) so file-plane API keys and
+  // env base URLs reach the gateway — a hand-rolled config here dropped them.
+  configureGateway(buildGatewayConfig(merged));
 
   // Read back resolved values — gateway applies internal defaults for unset
   // fields, so these are the values that actually shaped the schema.
@@ -831,13 +829,13 @@ async function initPGLite(opts: {
   // resolveAIOptions above: CLI flags > env vars > existing file > gateway
   // defaults.
   const { configureGateway } = await import('../core/ai/gateway.ts');
-  configureGateway({
+  configureGateway(buildGatewayConfig({
+    ...(loadConfig() ?? ({} as GBrainConfig)),
     embedding_model: resolvedModel ?? opts.aiOpts?.embedding_model,
     embedding_dimensions: resolvedDim ?? opts.aiOpts?.embedding_dimensions,
     expansion_model: opts.aiOpts?.expansion_model,
     chat_model: opts.aiOpts?.chat_model,
-    env: { ...process.env },
-  });
+  } as GBrainConfig));
   if (resolvedModel) console.log(`  Embedding: ${resolvedModel} (${resolvedDim}d)`);
   if (opts.aiOpts?.expansion_model) console.log(`  Expansion: ${opts.aiOpts.expansion_model}`);
   if (opts.aiOpts?.chat_model) console.log(`  Chat: ${opts.aiOpts.chat_model}`);
@@ -1046,13 +1044,13 @@ async function initPostgres(opts: {
 
   // T6: unconditional configureGateway BEFORE initSchema.
   const { configureGateway } = await import('../core/ai/gateway.ts');
-  configureGateway({
+  configureGateway(buildGatewayConfig({
+    ...(loadConfig() ?? ({} as GBrainConfig)),
     embedding_model: resolvedModel ?? opts.aiOpts?.embedding_model,
     embedding_dimensions: resolvedDim ?? opts.aiOpts?.embedding_dimensions,
     expansion_model: opts.aiOpts?.expansion_model,
     chat_model: opts.aiOpts?.chat_model,
-    env: { ...process.env },
-  });
+  } as GBrainConfig));
   if (resolvedModel) console.log(`  Embedding: ${resolvedModel} (${resolvedDim}d)`);
   if (opts.aiOpts?.expansion_model) console.log(`  Expansion: ${opts.aiOpts.expansion_model}`);
   if (opts.aiOpts?.chat_model) console.log(`  Chat: ${opts.aiOpts.chat_model}`);

src/commands/migrations/in-process.ts

@@ -66,15 +66,11 @@ export async function runMigrateOnlyCore(opts?: { timeoutMs?: number }): Promise
 
   // configureGateway BEFORE initSchema (init.ts B.3): a schema bump on a brain
   // whose file config is missing embedding fields must not fall through to
-  // stale hardcoded fallbacks. loadConfig already merged env; propagate it.
+  // stale hardcoded fallbacks. Route through buildGatewayConfig so file-plane
+  // API keys and provider base URLs follow the same precedence as runtime.
   const { configureGateway } = await import('../../core/ai/gateway.ts');
-  configureGateway({
-    embedding_model: config.embedding_model,
-    embedding_dimensions: config.embedding_dimensions,
-    expansion_model: config.expansion_model,
-    chat_model: config.chat_model,
-    env: { ...process.env },
-  });
+  const { buildGatewayConfig } = await import('../../core/ai/build-gateway-config.ts');
+  configureGateway(buildGatewayConfig(config));
 
   const timeoutMs = opts?.timeoutMs ?? MIGRATE_ONLY_TIMEOUT_MS;
   const engine = await createEngine(toEngineConfig(config));

src/core/ai/gateway.ts

@@ -683,33 +683,6 @@ export function getEmbeddingDimensions(): number {
   return requireConfig().embedding_dimensions ?? DEFAULT_EMBEDDING_DIMENSIONS;
 }
 
-/**
- * #2552: cap for parallel bulk-embed workers against a local inference
- * server. A single-slot Ollama/llama-server serializes requests, so the
- * cloud-tuned 20-worker fan-out multiplies latency x20 and blows past the
- * fetch timeout with no surfaced error (the backfill silently starves).
- */
-export const LOCAL_EMBED_CONCURRENCY_CAP = 2;
-
-/**
- * #2552: true when the configured embedding model routes to a local
- * inference server — the `ollama` / `llama-server` recipes, or any recipe
- * whose base URL was explicitly pointed at localhost. Bulk callers use this
- * to pick CPU-safe concurrency defaults; `gbrain doctor` uses it to warn
- * about an explicit cloud-sized override. Fail-open: unconfigured or
- * unresolvable gateway → false (cloud behavior, the historical default).
- */
-export function isLocalEmbeddingEndpoint(): boolean {
-  try {
-    const { recipe } = resolveRecipe(getEmbeddingModel());
-    if (recipe.id === 'ollama' || recipe.id === 'llama-server') return true;
-    const base = requireConfig().base_urls?.[recipe.id] ?? '';
-    return /\/\/(localhost|127\.0\.0\.1|\[::1\])(:|\/|$)/i.test(base);
-  } catch {
-    return false;
-  }
-}
-
 /**
  * v0.28.11: returns the configured multimodal embedding model when set,
  * or undefined if the brain falls back to `embedding_model` for multimodal

src/core/ai/recipes/ollama.ts

@@ -29,17 +29,9 @@ export const ollama: Recipe = {
       trust_custom_dims: true, // #2271: local models carry varied native dims
       cost_per_1m_tokens_usd: 0,
       price_last_verified: '2026-04-20',
-      // #2552: Ollama's true batch capacity depends on the locally loaded
-      // model + OLLAMA_NUM_PARALLEL, but the previous `no_batch_cap: true`
-      // meant a whole page went out in ONE request — on a CPU-only box that
-      // multiplies latency past the fetch timeout and the backfill starves
-      // with no surfaced error. Ollama doesn't return a recognizable
-      // token-limit error either, so the recursive-halving safety net never
-      // fires; a conservative static pre-split cap is the only guard.
-      // 4096 tokens x 2 chars/token ~= 8K chars per request (code-dense
-      // pages run ~2 chars/token, not the tiktoken-ish 4).
-      max_batch_tokens: 4096,
-      chars_per_token: 2,
+      // Ollama's batch capacity depends on the locally loaded model + the
+      // OLLAMA_NUM_PARALLEL config; no static cap to declare. v0.32 (#779).
+      no_batch_cap: true,
     },
   },
   setup_hint: 'Install Ollama from https://ollama.ai, then `ollama pull nomic-embed-text` and `ollama serve`.',

src/core/brain-score-recommendations.ts

@@ -12,8 +12,8 @@ import { parseModelId } from './ai/model-resolver.ts';
  * `resolveKey` closure without re-parsing recipes.
  *
  * Only OPENAI_API_KEY and ZEROENTROPY_API_KEY appear here because those are the
- * only embedding keys `buildGatewayConfig` (src/cli.ts) folds from config into
- * the gateway env. VOYAGE_API_KEY / GOOGLE_GENERATIVE_AI_API_KEY are deliberately
+ * only embedding keys `buildGatewayConfig` (src/core/ai/build-gateway-config.ts)
+ * folds from config into the gateway env. VOYAGE_API_KEY / GOOGLE_GENERATIVE_AI_API_KEY are deliberately
  * absent: their config fields are NOT threaded to the gateway today, so the
  * producer closures fall through to checking `process.env` ONLY for them. That
  * matches what the gateway can actually use (the recipes read those keys from

src/core/doctor-categories.ts

@@ -141,7 +141,6 @@ export const OPS_CHECK_NAMES: ReadonlySet<string> = new Set([
   'pgbouncer_prepare',
   'pgvector',
   'pool_budget',
-  'embed_concurrency',
   'progressive_batch_audit_health',
   'queue_health',
   'reranker_health',

src/core/pglite-engine.ts

@@ -56,7 +56,7 @@ import { GBrainError, PAGE_SORT_SQL, ENRICH_ORDER_SQL } from './types.ts';
 import { finalizeLastSeen } from './chronicle/last-seen.ts';
 import { computeAnomaliesFromBuckets } from './cycle/anomaly.ts';
 import { resolveBoostMap, resolveHardExcludes } from './search/source-boost.ts';
-import { buildSourceFactorCase, buildHardExcludeClause, buildVisibilityClause, buildRecencyComponentSql, buildBestPerPagePoolCte, buildOrFallbackWebsearchQuery, buildWebsearchQueryExpr } from './search/sql-ranking.ts';
+import { buildSourceFactorCase, buildHardExcludeClause, buildVisibilityClause, buildRecencyComponentSql, buildBestPerPagePoolCte, buildOrFallbackWebsearchQuery } from './search/sql-ranking.ts';
 import {
   normalizeEngineColumn,
   buildVectorCastFragment,
@@ -1591,8 +1591,6 @@ export class PGLiteEngine implements BrainEngine {
     }
 
     // v0.20.0 Cathedral II Layer 10 C1/C2: language + symbol-kind filters.
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query, innerLimit, limit, offset];
     let extraFilter = '';
     if (opts?.language) {
@@ -1632,7 +1630,6 @@ export class PGLiteEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
     const keywordSql =
       `WITH ranked AS (
@@ -1640,14 +1637,14 @@ export class PGLiteEngine implements BrainEngine {
            p.slug, p.id as page_id, p.title, p.type, p.source_id,
            p.effective_date, p.effective_date_source,
            cc.id as chunk_id, cc.chunk_index, cc.chunk_text, cc.chunk_source,
-           ts_rank(cc.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score,
+           ts_rank(cc.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score,
            CASE WHEN p.updated_at < (
              SELECT MAX(te.created_at) FROM timeline_entries te WHERE te.page_id = p.id
            ) THEN true ELSE false END AS stale
          FROM content_chunks cc
          JOIN pages p ON p.id = cc.page_id
          JOIN sources s ON s.id = p.source_id
-         WHERE cc.search_vector @@ ${ftsQueryExpr} ${detailFilter}${extraFilter} ${hardExcludeClause} ${visibilityClause}
+         WHERE cc.search_vector @@ websearch_to_tsquery('${ftsLang}', $1) ${detailFilter}${extraFilter} ${hardExcludeClause} ${visibilityClause}
            -- v0.27.1: hide image rows from default text-keyword search so
            -- OCR text doesn't drown text-page hits. Image-similarity queries
            -- run a separate vector path on embedding_image.
@@ -1715,10 +1712,7 @@ export class PGLiteEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query, limit, offset];
     let extraFilter = '';
     if (opts?.type) {
@@ -1766,7 +1760,7 @@ export class PGLiteEngine implements BrainEngine {
          COALESCE(rep.chunk_index, 0) as chunk_index,
          COALESCE(rep.chunk_text, '') as chunk_text,
          COALESCE(rep.chunk_source, 'compiled_truth') as chunk_source,
-         ts_rank_cd(p.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score,
+         ts_rank_cd(p.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score,
          CASE WHEN p.updated_at < (
            SELECT MAX(te.created_at) FROM timeline_entries te WHERE te.page_id = p.id
          ) THEN true ELSE false END AS stale
@@ -1781,7 +1775,7 @@ export class PGLiteEngine implements BrainEngine {
          ORDER BY (cc.chunk_source = 'compiled_truth') DESC, cc.chunk_index ASC
          LIMIT 1
        ) rep ON true
-       WHERE p.search_vector @@ ${ftsQueryExpr}
+       WHERE p.search_vector @@ websearch_to_tsquery('${ftsLang}', $1)
          ${extraFilter} ${hardExcludeClause} ${visibilityClause}
        ORDER BY score DESC, p.id ASC
        LIMIT $2 OFFSET $3`;
@@ -1968,8 +1962,6 @@ export class PGLiteEngine implements BrainEngine {
       });
     }
 
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query, limit, offset];
     let extraFilter = '';
     if (opts?.language) {
@@ -2004,21 +1996,20 @@ export class PGLiteEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
     const { rows } = await this.db.query(
       `SELECT
          p.slug, p.id as page_id, p.title, p.type, p.source_id,
          p.effective_date, p.effective_date_source,
          cc.id as chunk_id, cc.chunk_index, cc.chunk_text, cc.chunk_source,
-         ts_rank(cc.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score,
+         ts_rank(cc.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score,
          CASE WHEN p.updated_at < (
            SELECT MAX(te.created_at) FROM timeline_entries te WHERE te.page_id = p.id
          ) THEN true ELSE false END AS stale
        FROM content_chunks cc
        JOIN pages p ON p.id = cc.page_id
        JOIN sources s ON s.id = p.source_id
-       WHERE cc.search_vector @@ ${ftsQueryExpr} ${detailFilter}${extraFilter} ${hardExcludeClause} ${visibilityClause}
+       WHERE cc.search_vector @@ websearch_to_tsquery('${ftsLang}', $1) ${detailFilter}${extraFilter} ${hardExcludeClause} ${visibilityClause}
        ORDER BY score DESC
        LIMIT $2 OFFSET $3`,
       params

src/core/postgres-engine.ts

@@ -64,7 +64,7 @@ import { ConnectionManager } from './connection-manager.ts';
 import { logConnectionEvent } from './connection-audit.ts';
 import { validateSlug, contentHash, rowToPage, rowToStalePage, rowToChunk, rowToSearchResult, parseEmbedding, tryParseEmbedding, takeRowToTake, takeHitRowToHit, isUndefinedTableError, warnOncePerProcess } from './utils.ts';
 import { resolveBoostMap, resolveHardExcludes } from './search/source-boost.ts';
-import { buildSourceFactorCase, buildHardExcludeClause, buildVisibilityClause, buildRecencyComponentSql, buildBestPerPagePoolCte, buildOrFallbackWebsearchQuery, buildWebsearchQueryExpr } from './search/sql-ranking.ts';
+import { buildSourceFactorCase, buildHardExcludeClause, buildVisibilityClause, buildRecencyComponentSql, buildBestPerPagePoolCte, buildOrFallbackWebsearchQuery } from './search/sql-ranking.ts';
 import { DEFAULT_EMBEDDING_MODEL, DEFAULT_EMBEDDING_DIMENSIONS } from './ai/defaults.ts';
 import { DELETE_BATCH_SIZE } from './engine-constants.ts';
 
@@ -1691,8 +1691,6 @@ export class PostgresEngine implements BrainEngine {
     const hardExcludePrefixes = resolveHardExcludes(opts?.exclude_slug_prefixes, opts?.include_slug_prefixes);
     const hardExcludeClause = buildHardExcludeClause('p.slug', hardExcludePrefixes);
 
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query];
     let typeClause = '';
     if (type) {
@@ -1763,7 +1761,6 @@ export class PostgresEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
     const rawQuery = `
       WITH ranked_chunks AS (
@@ -1771,11 +1768,11 @@ export class PostgresEngine implements BrainEngine {
           p.slug, p.id as page_id, p.title, p.type, p.source_id,
           p.effective_date, p.effective_date_source,
           cc.id as chunk_id, cc.chunk_index, cc.chunk_text, cc.chunk_source,
-          ts_rank(cc.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score
+          ts_rank(cc.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score
         FROM content_chunks cc
         JOIN pages p ON p.id = cc.page_id
         JOIN sources s ON s.id = p.source_id
-        WHERE cc.search_vector @@ ${ftsQueryExpr}
+        WHERE cc.search_vector @@ websearch_to_tsquery('${ftsLang}', $1)
           ${typeClause}
           ${typesClause}
           ${excludeSlugsClause}
@@ -1866,10 +1863,7 @@ export class PostgresEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query];
     let typeClause = '';
     if (opts?.type) {
@@ -1929,7 +1923,7 @@ export class PostgresEngine implements BrainEngine {
         COALESCE(rep.chunk_index, 0) as chunk_index,
         COALESCE(rep.chunk_text, '') as chunk_text,
         COALESCE(rep.chunk_source, 'compiled_truth') as chunk_source,
-        ts_rank_cd(p.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score,
+        ts_rank_cd(p.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score,
         false AS stale
       FROM pages p
       JOIN sources s ON s.id = p.source_id
@@ -1942,7 +1936,7 @@ export class PostgresEngine implements BrainEngine {
         ORDER BY (cc.chunk_source = 'compiled_truth') DESC, cc.chunk_index ASC
         LIMIT 1
       ) rep ON true
-      WHERE p.search_vector @@ ${ftsQueryExpr}
+      WHERE p.search_vector @@ websearch_to_tsquery('${ftsLang}', $1)
         ${typeClause}
         ${typesClause}
         ${excludeSlugsClause}
@@ -2006,8 +2000,6 @@ export class PostgresEngine implements BrainEngine {
     const hardExcludePrefixes = resolveHardExcludes(opts?.exclude_slug_prefixes, opts?.include_slug_prefixes);
     const hardExcludeClause = buildHardExcludeClause('p.slug', hardExcludePrefixes);
 
-    // #2380: slash-bearing queries match both the split-word and literal
-    // slash forms — see buildWebsearchQueryExpr in ./search/sql-ranking.ts.
     const params: unknown[] = [query];
     let typeClause = '';
     if (type) {
@@ -2068,19 +2060,18 @@ export class PostgresEngine implements BrainEngine {
     // FTS config name (e.g. 'english', 'pt_br'). Validated by getFtsLanguage()
     // — safe to interpolate into raw SQL.
     const ftsLang = getFtsLanguage();
-    const ftsQueryExpr = buildWebsearchQueryExpr(ftsLang, '$1', query);
 
     const rawQuery = `
       SELECT
         p.slug, p.id as page_id, p.title, p.type, p.source_id,
         p.effective_date, p.effective_date_source,
         cc.id as chunk_id, cc.chunk_index, cc.chunk_text, cc.chunk_source,
-        ts_rank(cc.search_vector, ${ftsQueryExpr}) * ${sourceFactorCase} AS score,
+        ts_rank(cc.search_vector, websearch_to_tsquery('${ftsLang}', $1)) * ${sourceFactorCase} AS score,
         false AS stale
       FROM content_chunks cc
       JOIN pages p ON p.id = cc.page_id
       JOIN sources s ON s.id = p.source_id
-      WHERE cc.search_vector @@ ${ftsQueryExpr}
+      WHERE cc.search_vector @@ websearch_to_tsquery('${ftsLang}', $1)
         ${typeClause}
         ${typesClause}
         ${excludeSlugsClause}

src/core/search/sql-ranking.ts

@@ -251,28 +251,6 @@ export function buildOrFallbackWebsearchQuery(query: string): string | null {
   return tokens.join(' OR ');
 }
 
-/**
- * #2380: FTS query expression for slash-bearing queries. Postgres' default
- * text-search parser classifies `foo/bar` as a single `file`-alias lexeme —
- * on BOTH the query side and the index side. So a raw `foo/bar` query only
- * matched documents carrying the identical joined lexeme (literal paths),
- * and a slash-split query only matches documents whose text had the words
- * separated. Neither form alone covers both document shapes; OR the two
- * parses so a slash query matches prose ("foo and bar", stemmed, AND
- * semantics) AND literal slash forms ("src/core/x.ts") alike.
- *
- * Slash-free queries return the plain single-parse expression — byte-
- * identical SQL and identical ts_rank to the historical behavior.
- *
- * `ftsLang` is validated by getFtsLanguage() (safe to interpolate);
- * `param` is a `$N` placeholder, never user text.
- */
-export function buildWebsearchQueryExpr(ftsLang: string, param: string, query: string): string {
-  const plain = `websearch_to_tsquery('${ftsLang}', ${param})`;
-  if (!query.includes('/')) return plain;
-  return `(websearch_to_tsquery('${ftsLang}', translate(${param}, '/', ' ')) || ${plain})`;
-}
-
 // ============================================================
 // v0.29.1 — Recency component SQL builder
 // ============================================================

test/ai/build-gateway-config.test.ts

@@ -25,6 +25,7 @@ import { withEnv } from '../helpers/with-env.ts';
 
 const PASSTHROUGHS: Array<{ envVar: string; recipeId: string }> = [
   { envVar: 'LLAMA_SERVER_BASE_URL', recipeId: 'llama-server' },
+  { envVar: 'LLAMA_SERVER_RERANKER_BASE_URL', recipeId: 'llama-server-reranker' },
   { envVar: 'OLLAMA_BASE_URL', recipeId: 'ollama' },
   { envVar: 'LMSTUDIO_BASE_URL', recipeId: 'lmstudio' },
   { envVar: 'LITELLM_BASE_URL', recipeId: 'litellm' },

test/ai/no-batch-cap-suppression.serial.test.ts

@@ -28,8 +28,8 @@ describe('v0.32 #779: no_batch_cap suppresses the missing-max_batch_tokens warni
     resetGateway();
   });
 
-  test('LiteLLM and llama-server declare no_batch_cap: true', () => {
-    for (const id of ['litellm', 'llama-server']) {
+  test('Ollama, LiteLLM, llama-server all declare no_batch_cap: true', () => {
+    for (const id of ['ollama', 'litellm', 'llama-server']) {
       const r = getRecipe(id);
       expect(r, `${id} not registered`).toBeDefined();
       expect(
@@ -39,18 +39,6 @@ describe('v0.32 #779: no_batch_cap suppresses the missing-max_batch_tokens warni
     }
   });
 
-  test('#2552: Ollama declares a conservative static batch cap, not no_batch_cap', () => {
-    // A CPU-only Ollama box wedges when a whole page ships in one request;
-    // Ollama never returns a token-limit error so the recursive-halving
-    // safety net can't fire. The pre-split cap is the only guard.
-    const r = getRecipe('ollama');
-    expect(r).toBeDefined();
-    const e = r!.touchpoints.embedding!;
-    expect(e.no_batch_cap).toBeUndefined();
-    expect(e.max_batch_tokens).toBe(4096);
-    expect(e.chars_per_token).toBe(2);
-  });
-
   test('configureGateway does NOT warn for ollama/litellm/llama-server', () => {
     warnSpy.mockClear();
     resetGateway();

test/cli-multimodal-integration.test.ts

@@ -11,32 +11,13 @@
 import { afterAll, beforeAll, beforeEach, describe, expect, test } from 'bun:test';
 import { PGLiteEngine } from '../src/core/pglite-engine.ts';
 import { loadConfigWithEngine, type GBrainConfig } from '../src/core/config.ts';
+import { buildGatewayConfig } from '../src/core/ai/build-gateway-config.ts';
 import {
   configureGateway,
   getEmbeddingModel,
   getMultimodalModel,
   resetGateway,
 } from '../src/core/ai/gateway.ts';
-import type { AIGatewayConfig } from '../src/core/ai/types.ts';
-
-// Mirror the cli.ts buildGatewayConfig helper exactly. Keeping a copy here
-// (instead of exporting from cli.ts) is intentional: the test asserts the
-// shape of the contract, not the helper's identity. If cli.ts drifts, the
-// e2e behavior these tests care about (DB-set value lands in gateway) still
-// holds, but a helper-shape test would also catch the drift in PR review.
-function buildGatewayConfig(c: GBrainConfig): AIGatewayConfig {
-  return {
-    embedding_model: c.embedding_model,
-    embedding_dimensions: c.embedding_dimensions,
-    embedding_multimodal_model: c.embedding_multimodal_model,
-    expansion_model: c.expansion_model,
-    chat_model: c.chat_model,
-    chat_fallback_chain: c.chat_fallback_chain,
-    base_urls: c.provider_base_urls,
-    provider_chat_options: c.provider_chat_options,
-    env: { ...process.env },
-  };
-}
 
 let engine: PGLiteEngine;
 
@@ -47,6 +28,7 @@ beforeAll(async () => {
 });
 
 afterAll(async () => {
+  resetGateway(); // don't leak this file's gateway config into shard siblings
   await engine.disconnect();
 });
 

test/embed-local-endpoint.serial.test.ts

@@ -1,118 +0,0 @@
-/**
- * #2552: cloud-tuned embedding defaults silently wedge CPU-only local
- * endpoints (Ollama). Three-part fix under test:
- *
- *  1. `isLocalEmbeddingEndpoint()` — gateway helper detecting local
- *     inference servers (ollama / llama-server recipes, localhost base URL).
- *  2. `resolveEmbedConcurrency()` — embed auto-caps the 20-worker fan-out
- *     at LOCAL_EMBED_CONCURRENCY_CAP for local endpoints unless the
- *     operator set GBRAIN_EMBED_CONCURRENCY explicitly.
- *  3. `computeEmbedConcurrencyCheck()` — doctor warns when an explicit env
- *     override fans out against a local endpoint.
- *
- * Serial: mutates process.env and the module-global gateway config.
- */
-
-import { afterAll, afterEach, describe, expect, test } from 'bun:test';
-import {
-  configureGateway,
-  resetGateway,
-  isLocalEmbeddingEndpoint,
-  LOCAL_EMBED_CONCURRENCY_CAP,
-} from '../src/core/ai/gateway.ts';
-import { resolveEmbedConcurrency } from '../src/commands/embed.ts';
-import { computeEmbedConcurrencyCheck } from '../src/commands/doctor.ts';
-
-const SAVED_ENV = process.env.GBRAIN_EMBED_CONCURRENCY;
-
-afterEach(() => {
-  resetGateway();
-  if (SAVED_ENV === undefined) delete process.env.GBRAIN_EMBED_CONCURRENCY;
-  else process.env.GBRAIN_EMBED_CONCURRENCY = SAVED_ENV;
-});
-
-afterAll(() => {
-  resetGateway();
-});
-
-describe('#2552 isLocalEmbeddingEndpoint', () => {
-  test('false when the gateway is not configured (fail-open to cloud behavior)', () => {
-    resetGateway();
-    expect(isLocalEmbeddingEndpoint()).toBe(false);
-  });
-
-  test('true for the ollama recipe', () => {
-    configureGateway({ embedding_model: 'ollama:nomic-embed-text', env: {} });
-    expect(isLocalEmbeddingEndpoint()).toBe(true);
-  });
-
-  test('true for the llama-server recipe', () => {
-    configureGateway({ embedding_model: 'llama-server:my-gguf', env: {} });
-    expect(isLocalEmbeddingEndpoint()).toBe(true);
-  });
-
-  test('false for a cloud recipe', () => {
-    configureGateway({
-      embedding_model: 'openai:text-embedding-3-small',
-      env: { OPENAI_API_KEY: 'fake' },
-    });
-    expect(isLocalEmbeddingEndpoint()).toBe(false);
-  });
-
-  test('true when a cloud recipe base URL is explicitly pointed at localhost', () => {
-    configureGateway({
-      embedding_model: 'openai:text-embedding-3-small',
-      env: { OPENAI_API_KEY: 'fake' },
-      base_urls: { openai: 'http://localhost:8080/v1' },
-    });
-    expect(isLocalEmbeddingEndpoint()).toBe(true);
-  });
-});
-
-describe('#2552 resolveEmbedConcurrency', () => {
-  test('caps at LOCAL_EMBED_CONCURRENCY_CAP for a local endpoint when env is unset', () => {
-    delete process.env.GBRAIN_EMBED_CONCURRENCY;
-    configureGateway({ embedding_model: 'ollama:nomic-embed-text', env: {} });
-    expect(resolveEmbedConcurrency()).toBe(LOCAL_EMBED_CONCURRENCY_CAP);
-  });
-
-  test('explicit env override always wins, even against a local endpoint', () => {
-    process.env.GBRAIN_EMBED_CONCURRENCY = '10';
-    configureGateway({ embedding_model: 'ollama:nomic-embed-text', env: {} });
-    expect(resolveEmbedConcurrency()).toBe(10);
-  });
-
-  test('cloud endpoints keep the historical default of 20', () => {
-    delete process.env.GBRAIN_EMBED_CONCURRENCY;
-    configureGateway({ env: { OPENAI_API_KEY: 'fake' } });
-    expect(resolveEmbedConcurrency()).toBe(20);
-  });
-
-  test('pacing only ever lowers concurrency', () => {
-    delete process.env.GBRAIN_EMBED_CONCURRENCY;
-    configureGateway({ embedding_model: 'ollama:nomic-embed-text', env: {} });
-    expect(resolveEmbedConcurrency(1)).toBe(1);
-    expect(resolveEmbedConcurrency(16)).toBe(LOCAL_EMBED_CONCURRENCY_CAP);
-  });
-});
-
-describe('#2552 computeEmbedConcurrencyCheck (doctor)', () => {
-  test('ok for non-local endpoints', () => {
-    expect(computeEmbedConcurrencyCheck(false, '20', 2).status).toBe('ok');
-  });
-
-  test('warn when an explicit override exceeds the local cap', () => {
-    const check = computeEmbedConcurrencyCheck(true, '20', 2);
-    expect(check.status).toBe('warn');
-    expect(check.message).toContain('GBRAIN_EMBED_CONCURRENCY=20');
-  });
-
-  test('ok when env is unset against a local endpoint (auto-cap applies)', () => {
-    expect(computeEmbedConcurrencyCheck(true, undefined, 2).status).toBe('ok');
-  });
-
-  test('ok when the override is at or under the cap', () => {
-    expect(computeEmbedConcurrencyCheck(true, '2', 2).status).toBe('ok');
-    expect(computeEmbedConcurrencyCheck(true, '1', 2).status).toBe('ok');
-  });
-});

test/eval-takes-quality-gateway.test.ts

@@ -0,0 +1,57 @@
+/**
+ * eval-takes-quality gateway self-config — adapter-boundary regression
+ * (takeover of PR #2430).
+ *
+ * The old callsite spread `{ ...cfg, ...process.env }` straight into
+ * configureGateway. The gateway NEVER reads process.env at call time — it
+ * reads `_config.env` — and that spread never populated an `env` field at
+ * all, so every availability/diagnose check dereferenced `undefined.env[k]`
+ * and file-plane API keys (config.json `openai_api_key` etc.) were dropped.
+ * Routing through buildGatewayConfig fixes both. This test fails (throws)
+ * on the old code path.
+ */
+import { afterAll, describe, expect, test } from 'bun:test';
+import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { runEvalTakesQuality } from '../src/commands/eval-takes-quality.ts';
+import { isAvailable, resetGateway } from '../src/core/ai/gateway.ts';
+import { withEnv } from './helpers/with-env.ts';
+import type { BrainEngine } from '../src/core/engine.ts';
+
+afterAll(() => {
+  resetGateway();
+});
+
+describe('runEvalTakesQuality — gateway self-config routes through buildGatewayConfig', () => {
+  test('file-plane openai_api_key reaches the gateway env (help path, engine untouched)', async () => {
+    const home = mkdtempSync(join(tmpdir(), 'gbrain-etq-gw-'));
+    try {
+      mkdirSync(join(home, '.gbrain'), { recursive: true });
+      writeFileSync(
+        join(home, '.gbrain', 'config.json'),
+        JSON.stringify({
+          engine: 'pglite',
+          database_path: join(home, '.gbrain', 'brain'),
+          openai_api_key: 'sk-file-plane-test',
+        }),
+      );
+      await withEnv(
+        {
+          GBRAIN_HOME: home,
+          OPENAI_API_KEY: undefined,
+          DATABASE_URL: undefined,
+          GBRAIN_DATABASE_URL: undefined,
+        },
+        async () => {
+          // 'help' returns before touching the engine, but the gateway is
+          // configured first — exactly the seam under test.
+          await runEvalTakesQuality({} as BrainEngine, ['--help']);
+          expect(isAvailable('embedding', 'openai:text-embedding-3-small')).toBe(true);
+        },
+      );
+    } finally {
+      rmSync(home, { recursive: true, force: true });
+    }
+  });
+});

test/pglite-engine.test.ts

@@ -216,67 +216,6 @@ describe('PGLiteEngine: Search', () => {
     expect(results.length).toBe(0);
   });
 
-  // Regression (#2380): queries containing `/` used to bypass FTS AND
-  // semantics. Postgres' default text-search parser classifies `foo/bar` as
-  // a `file`-alias token mapped to the `simple` dictionary, so it became a
-  // single un-stemmed lexeme `'foo/bar'` that never matches indexed text —
-  // the primary FTS pass returned 0 and the OR fallback took over, matching
-  // pages that contain EITHER term. searchKeyword/searchTitles now normalize
-  // `/` to whitespace before websearch_to_tsquery parses, so the primary
-  // AND pass matches directly.
-  test('searchKeyword: slash query matches with AND semantics, not OR fallback', async () => {
-    // Decoy shares only ONE of the two query terms ('enterprise').
-    await engine.putPage('concepts/enterprise-pricing', {
-      type: 'concept', title: 'Widget Pricing',
-      compiled_truth: 'Enterprise pricing for widgets.',
-    });
-    await engine.upsertChunks('concepts/enterprise-pricing', [
-      { chunk_index: 0, chunk_text: 'Enterprise pricing for widgets', chunk_source: 'compiled_truth' },
-    ]);
-
-    // Both terms co-occur only in the novamind chunk. Pre-fix this returned
-    // BOTH pages (primary pass zero-hit → OR fallback); post-fix the primary
-    // AND pass returns exactly the co-occurrence page.
-    const results = await engine.searchKeyword('NovaMind/enterprise');
-    expect(results.length).toBe(1);
-    expect(results[0].slug).toBe('companies/novamind');
-  });
-
-  test('searchTitles: slash query matches with AND semantics, not OR fallback', async () => {
-    await engine.putPage('companies/novamind-enterprise', {
-      type: 'company', title: 'NovaMind Enterprise Platform',
-      compiled_truth: 'Placeholder body.',
-    });
-    await engine.putPage('guides/enterprise-sales', {
-      type: 'concept', title: 'Enterprise Sales Guide',
-      compiled_truth: 'Placeholder body.',
-    });
-
-    // Pre-fix: `NovaMind/Enterprise` parsed as one file-alias lexeme → the
-    // primary title pass returned 0 and the OR fallback matched BOTH titles.
-    const results = await engine.searchTitles('NovaMind/Enterprise');
-    expect(results.length).toBe(1);
-    expect(results[0].slug).toBe('companies/novamind-enterprise');
-  });
-
-  test('searchKeyword: slash query still matches the literal slash form (file paths)', async () => {
-    // The INDEX side also emits the joined file-alias lexeme for literal
-    // `foo/bar` text, so a query normalized to split words alone would go
-    // blind to documents containing the literal slash form (paths, URLs).
-    // buildWebsearchQueryExpr ORs both parses; this pins the raw arm.
-    await engine.putPage('runbooks/widget-deploy', {
-      type: 'concept', title: 'Widget Deploy Runbook',
-      compiled_truth: 'Runbook for the acme/widget deployment pipeline.',
-    });
-    await engine.upsertChunks('runbooks/widget-deploy', [
-      { chunk_index: 0, chunk_text: 'Runbook for the acme/widget deployment pipeline', chunk_source: 'compiled_truth' },
-    ]);
-
-    const results = await engine.searchKeyword('acme/widget');
-    expect(results.length).toBe(1);
-    expect(results[0].slug).toBe('runbooks/widget-deploy');
-  });
-
   test('tsvector trigger populates search_vector on insert', async () => {
     // Verify the PL/pgSQL trigger fires and content_chunks.search_vector is
     // populated from chunk_text. v0.20.0 Cathedral II Layer 3 moved FTS from