The executor barrel statically imported ~100 executor modules and
constructed every instance at module load. Measured cold cost on top of
the minimal set: ~0.7–1.2s boot time and ~35MB heap, paid by every
deployment regardless of which providers it uses.
Now:
- executors/index.ts keeps the declarative alias table byte-stable (same
keys, same order, same ctor args — pinned by the golden lock) but each
value is a deferred loader using dynamic import; bundlers emit
on-demand chunks
- registry.ts gains registerLazyExecutor/loadRegisteredExecutor: aliases
are declared eagerly so hasSpecializedExecutor() and
listExecutorAliases() stay synchronous, instances materialize once on
first use and cache into the same registry map
- getExecutor() becomes async; production call sites (chatCore proxy
resolver, video generation, compression judge/eval clients,
quotaAutoPing deps, anthropic OAuth validation) await it
- cliproxy wrapper ExecutorLike types drop their index signatures so
BaseExecutor satisfies them structurally
Measured after (isolated DATA_DIR): barrel boot 712-832ms / ~45MB with
first-use materialization of an executor costing +120-150ms once.
Test impact: 24 unit suites adapted mechanically to the async seam
(await + union narrowing on the Response | {response} execute result);
class imports moved from the barrel to executor module files. The
web-cookie sweep SIGABRT failure is pre-existing (reproduced identically
on the clean base).
Commit gate note: husky lint-staged fails with 'suppressions left that
do not occur anymore' — reproduced identically on a stashed clean tree
(22 baseline problems), independent of this change.
Our PR fixed the last no-restricted-imports violation in
src/lib/memory/embedding/index.ts (localDb barrel import replaced by a
direct readCache import), so the frozen suppression entry for that file
became stale. The gate (run-eslint-json.mjs) hard-fails on stale entries,
so remove it to keep 'No new ESLint warnings' green.
The memory Engine tab quick-select built its model list only from a
keyword heuristic over the chat catalog plus OpenRouter live discovery,
so configured providers whose embedding models are not in that catalog
(e.g. Cloudflare Workers AI cf/@cf/baai/bge-m3) never appeared, and
requests for them failed with 'Unknown embedding provider'.
- add cloudflare-ai to EMBEDDING_PROVIDERS with a requiresAccountId URL
template resolved per-request via buildEmbeddingProviderUrl() (mirrors
CloudflareAIExecutor on the chat side) plus the cf provider alias
- merge curated registry models into the quick-select catalog and add a
generic fallback listing: any configured OpenAI-compatible chat
provider without a curated entry appears with free-text model input
- same treatment for rerank: listRerankProviders + generic Cohere-
compatible fallback, new /api/memory/rerank-providers endpoint,
selectors fall back to free-text when no static models exist
- runtime: embeddings service/handler and rerank route/handler resolve
derived providers instead of rejecting unlisted ones
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Resolvido o mesmo conflito não-relacionado em src/shared/utils/wsPath.ts (mesma causa do #11436 — refactor já mergeado na branch depois do fork deste PR; o diff real deste PR — bin/cli/utils/volatileEnvPath.mjs + bin/omniroute.mjs — ficou intacto) e revalidado: typecheck:core limpo, 12/12 testes focados passando.
Companion do #11436, decisão pura testável isoladamente, sem mudança de comportamento fora do caso volátil. Obrigado pela contribuição!
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Resolvido um conflito de merge não-relacionado em src/shared/utils/wsPath.ts (originado de um refactor já mergeado nessa branch depois do fork deste PR; o diff real deste PR — scripts/dev/sync-env.mjs + tests/unit/sync-env.test.ts — ficou intacto) e revalidado: typecheck:core limpo, 13/13 testes focados passando.
Segue o precedente correto do #1622 (STORAGE_ENCRYPTION_KEY) para os dois secrets restantes que a postinstall preenchia por engano, defeituando o mecanismo de ensureSecrets(). Obrigado pela contribuição!
The cycle-open commit's root CHANGELOG hunk was lost in the rebase onto the
branch tip (the 42 i18n mirrors kept theirs, so the root and the mirrors had
drifted apart). Re-inserts the section with the three canonical headings.
Completes the 0a.0b cut for the parallel-cycle model: the branch already existed
(cut from the v3.8.50 tip) but had never been bumped. Bumps package.json x3,
openapi.yaml and the lockfile, adds the living [3.8.51] CHANGELOG section with
the three canonical headings so aggregate-changelog.mjs cannot mis-target an
older published section, and syncs the 42 i18n mirrors.
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Validado em lote combinado (batch-0824h2) contra o tip de release/v3.8.51: typecheck:core limpo, gates estáticos + migration-numbering OK, 127/127 testes focados passando (8/8 do PR entre migration-163 e radar-feed-cache-generated-at).
Migração limpa (ADD COLUMN nullable, sem backfill necessário), aditiva na API, mantém "unknown" honesto para linhas antigas. Obrigado pela contribuição!
Retargetado para release/v3.8.51 (release/v3.8.50 está congelada — freeze issue #11439). Validado em lote combinado (batch-0824h2, junto de #11435/#11436/#11437) contra o tip de release/v3.8.51: typecheck:core limpo, gates estáticos OK, 127/127 testes focados passando.
Investigação sólida com repro real via container isolado, três causas independentes identificadas e corrigidas com testes de regressão dedicados para cada uma. Obrigado pela contribuição!
Grounded design answering the five open questions in #8075 with a
measured coupling census of release/v3.8.51: 126 db modules on the sync
prepare()/transaction() shape, 36 pragma sites, 159 SQLite migrations,
9 sqlite-vec + 8 FTS5 files, and a 4-driver SqliteAdapter seam that is
SQLite-only in practice (synchronous API).
Proposal: dialect-agnostic async DurableStateStore port with per-domain
repositories adopted incrementally; Postgres-first with MySQL gated by
a shared dialect-conformance suite; DATABASE_DRIVER config surface with
sqlite as the non-negotiable zero-config default; fail-closed one-way
export tool; five-phase delivery plan.
Dispatch: OpencodeExecutor.transformRequest stripped the effort-suffixed
alias (e.g. glm-5.2-high) down to the base id and injected a flat
reasoning_effort body field — but opencode-go's ChatCompletionRequest has
no such field for non-DeepSeek families, so the tier never reached the
upstream and every request ran at default effort. Only DeepSeek V4
accepts the flat field (its native contract, #4647), so it keeps the old
rewrite; every other family now forwards the aliased model id verbatim,
which is their only native effort mechanism.
Registry: declare supportedThinkingEfforts on the base rows that had
aliases but no tier vocabulary — opencode-go (glm-5.2, mimo-v2.5,
grok-4.5, hy3, kimi-k3, qwen3.7-max/plus), the shared zen/go row
(qwen3.6-plus) and opencode-zen (deepseek-v4-pro/flash, glm-5.2,
kimi-k3) — so catalog variant synthesis (#9485) and
sanitizeReasoningEffortForProvider clamp from one source of truth.
nvidia z-ai/glm-5.2 is declared reasoning-capable with an EMPTY tier
list: its upstream only exposes a binary enable_thinking switch
(mapNvidiaGlm52ReasoningParams), so no honest tiers exist to advertise.
Ollama Cloud declarations were already landed upstream (#11307); this
change covers the remaining providers named in #10788.
Merging --admin with red discrimination (merge-gates §4). 5/6 failing tests are the compression/i18n-vi/kiro/memory mini-cluster fixed by #11306 (this branch predates it). The 6th (token-health-check-kimi, 'jittered expiration window') is a timing-sensitive test — reproduces GREEN locally, name itself indicates randomized jitter, not a deterministic regression from this PR's reasoning-effort clamp change (zero file overlap). ESLint warnings + dast-smoke (advisory, isRequired:null) also inherited.
Merging --admin: this is itself the base-red drain PR (compression/i18n-vi/i18n-pt-br/kiro/memory mini-cluster). Remaining fails (ESLint warnings ratchet drift, dast-smoke advisory) are not blocking; isRequired:null on dast-smoke.
Validated on a 4-PR combined board: claude-classifier-compat 15/15 + 8189-classifier-compat-auto-narrow 5/5 sibling sweep, typecheck:core clean, gates within baseline. Detects the newer Claude Code severity-classifier request shape (stop_sequences carrying </severity>) and replies with <severity>0</severity> instead of the legacy <block>no</block> — the mismatch was failing both classifier stages closed and blocking gated tool calls (Bash, WebSearch). Closes#11289.
Validated on a 4-PR combined board: memory-cache-safe-injection + memory-glm-injection + chatcore-memory-skills-injection 39/39 (node:test), injection.test.ts 23/23 (vitest), typecheck:core clean, check:cycles clean, gates within baseline. Narrows the #3890 mid-array system-splice skip to the exact shape that 400s on Claude Opus 5 (splice point not preceded by a server tool result) instead of disabling the cache-safe optimization for every Claude request — preserves the fast path everywhere it's actually safe. Closes#11290.
Validated on a 4-PR combined board: provider-patch-ratelimit-protection-11278 2/2 + 56/56 sibling suites, typecheck:core clean, gates within baseline. PUT /api/providers/[id] can no longer silently enable rate-limit protection just because EditConnectionModal sends rateLimitOverrides on every save — the runtime toggle now strictly follows the persisted DB row. Closes#11278.
Validated on a 4-PR combined board: @omniroute/opencode-plugin provider.test.ts 13/13, full plugin suite 343/343, YAML validated, typecheck:core clean, gates within baseline. CI branch trigger was pinned to release/v3.8.2 since creation (never fired mid-cycle) — now tracks [main, release/**] matching every other workflow's pattern; combo-id fixture aligned to the real #10821 unprefixed-lookup contract (RED 11/13 → GREEN 13/13).
Validated on a 3-PR combined board: openai-responses-opencode-subagent-sessionid within the board's 20/20, typecheck:core + check:open-sse-typecheck clean, gates within baseline. Uses null as the omission sentinel for OpenCode's optional default-less subagent.sessionID under Responses/Codex strict mode (which required every declared property) — models can now omit the field to spawn a new child instead of inventing filler IDs; the null is stripped before OpenCode sees the call, real child IDs pass through unchanged. Thank you @ofonseca-pyming!
Validated on a 3-PR combined board: openapi-security-tiers 1/1 within the board's 20/20, typecheck:core clean, gates within baseline. Declares GET /api/openapi/spec's real ManagementSessionAuth contract (conditional on requireLogin) — no runtime behavior change, just the doc catching up. Closes#10851. Thank you @RaviTharuma!
Validated on a 3-PR combined board: 11018-database-cache-docs 1/1 within the board's 20/20 focused suite, typecheck:core clean, gates within baseline. DATABASE_GUIDE.md now cites the real 64 MiB runtime default and the 1..1,000,000 KiB range, locked with a regression test. Thank you @RaviTharuma!
Validated on a 2-PR combined board: 158/158 across the drain's 17 focused test files (incl. integration qdrant-routes), typecheck:core + dashboard-typecheck clean, env-doc-sync, mutation-test-coverage, cli-i18n, file-size, changelog-integrity all green. Every red discriminated per-item as stale-test (contract intentionally moved, citing the origin PR) or real bug (fixed, never masked) across 3 base-advance waves (#9985): stryker registration, i18n gaps (health.* namespace mismatch + pt-BR parity), combo-token-window pins (872K), CLI catalog counts (prime-agent), the isComboStep single-target reasoning-transport-fallback regression, container-guard hermeticity, and 26 unsuppressed ESLint errors from the wave (typed qdrant adapter, import trim, justified disable with precedent). Drains the accumulated base-red cluster.
Validated on a 2-PR combined board: routeGuard 36/36 (within the 68/68 focused-file total), a2a-task-owner-idor 7/7, a2a-tasks-auth, search-baseurl-ssrf-guard, cli-serve-hostname, spawn-capable-prefixes-client-safe all green, typecheck:core + dashboard-typecheck clean, gates within baseline. Five real High-severity advisories fixed with TDD (each failing-then-passing): settings export/import-json ALWAYS_PROTECTED completion, MITM route LOCAL_ONLY+SPAWN_CAPABLE gating, search baseUrl SSRF/IMDS guard, A2A REST task auth+ownership (previously none at all), and the loud boot exposure warning. GHSA-cjv9 confirmed already closed on this base (verified). Round 3 of the advisory sweep.
Validated on a 3-PR combined board: traffic-inspector-beginner-header suite green within the board's 30/30, typecheck:core + dashboard-typecheck clean, gates within baseline. Purpose-first orientation header for Traffic Inspector, existing inspection UI untouched. Thank you @ignamiranda!
Validated on a 3-PR combined board: sidebar-customization + sidebar-essentials-static 30/30 combined (across the board's 4 focused files), typecheck:core + dashboard-typecheck clean, gates within baseline. New Essentials sidebar preset gives first-time users a short beginner path while advanced tools stay reachable via Command Palette search. Thank you @ignamiranda!
Validated on a 3-PR combined board: kiro-connection-identity 8/8 (written failing-first — 3 new cases red on the pristine release/v3.8.50 tip, green with this change), typecheck:core + dashboard-typecheck clean, all static gates within baseline. Root cause is exactly right: a CodeWhisperer profile ARN identifies the profile, not the account, and distinct Builder ID accounts via social login can share one — the ARN is now trusted only alongside a non-contradicting account-level identifier (email/clientId). Closes#10815. Thank you @engenhariaandrereis01-ai!
- open-sse/executors/github.ts: replace the Math.random() fallback in
the Copilot correlation-id generators (x-request-id,
x-interaction-id, x-client-session-id, x-agent-task-id) with a
CSPRNG-backed randomIdFallback() (node:crypto randomBytes) — closes
js/insecure-randomness with no behavior change (crypto.randomUUID
stays the primary path).
- tests/unit/cli/_helpers/shellArgs.mjs: collapse the two sequential
global .replace() unescape passes into a single left-to-right regex
replace with alternation — closes js/double-escaping. The prior
two-pass form let the first pass's output feed the second, which is
exactly the double-(un)escaping bug pattern the query flags (e.g. an
escaped-backslash-then-quote sequence could be misread depending on
pass order).
Co-authored-by: Markus Hartung <mail@hartmark.se>
Validated on the combined 8-PR board + the branch itself: vertex-anthropic-models 4/4 (new — pushed to your branch as a fix-in-place commit per Hard Rule #8, covering the parser's global/project-scoped resource names, malformed-input handling, and the claude-* → targetFormat heuristic on vertex/vertex-partner), 88/88 across the board's focused suites, typecheck:core clean, gates within baseline. Anthropic partner models on Vertex AI now discover dynamically via the Model Garden publisher endpoint and route through the Claude translator even for models outside the static registry. Thank you @maci0!
Validated on the combined 8-PR board: all seven CLI command suites green (combo-suggest, compression, mcp-call, oneproxy, resilience, skills + the shared mcpStreamMock helper) within the board's 88/88, typecheck:core clean, gates within baseline. The CLI mcp call protocol fixes (BUG-001) land with full regression coverage across the affected commands. Thank you @YunyunZhai — and thank you for your patience while this one waited for review.
