maint: final follow-up cherry-pick #9783 (#9904)

* fix(deps): bump transitive deps for 6 Dependabot + remaining audit vulns on main

Same overrides as #9464 (ip-address, hono, fast-uri, socket.io-parser, undici)
applied directly to main. Also covers brace-expansion (scoped), js-yaml v4 copies,
and mermaid.

npm audit: 6→0 vulnerabilities.
Closes Dependabot #161-#166.

* fix(deps): bump nanoid, dompurify for 2 new Dependabot alerts (#189, #190)

Bumps: nanoid ^3.3.17 (was transitive, now overridden), dompurify ^3.4.13
(with monaco-editor scoped override). Closes Dependabot #189, #190.

Remaining #182-#188 (js-yaml + mermaid) already closed by #9651 merge —
awaiting Dependabot re-scan.

npm audit → 0 vulnerabilities.

* fix(repo): harden .gitignore to also ignore a _tasks symlink (/_tasks)

_tasks is a SEPARATE nested git repo (gitignored). The pattern _tasks/ (trailing
slash) ignores only a directory, not a SYMLINK named _tasks. A self-referential
_tasks symlink can slip in via git add -A and, once pulled, checkout materializes
it over the real _tasks repo (destroying plans/specs/hands-off). Anchored /_tasks
ignores the symlink too, preventing re-capture.

* fix(translator): keep Responses namespace identity across the hub-and-spoke pivot

Step 1 of the pivot (openai-responses -> openai) flattens namespace sub-tools
to a qualified wire name (#8295) and records the `{namespace, name}` pair on a
non-enumerable `_toolNameMap`. Step 2 (openai -> target) returns a brand-new
object, so the property was dropped for every non-OpenAI target. chatCore then
handed `null` to the #7936 response seam and namespace sub-tool calls reached
the client under their flattened name, which Codex rejects with
`unsupported call: <name>` — the symptom #7936 was opened to fix.

Copying `_toolNameMap` through is not viable: openai-to-claude and
openai-to-gemini publish their own `Map<string, string>` alias map on that same
property during step 2, so it carries two incompatible types. This adds a
dedicated `_namespaceToolIdentityMap`, propagated by translateRequest across
the pivot; chatCore prefers it and falls back to `_toolNameMap` for the
non-pivot producers. Both keys are stripped from the cliproxyapi wire body.

Fixes #9780

* fix(chat): reduce file size

Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>

* fix(chat): reduce combined file size

Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>

* fix(chat): reduce combined file size

Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>

---------

Co-authored-by: diegosouzapw <diegosouzapw@users.noreply.github.com>
Co-authored-by: VXNCXNX <vincent@preuve.ai>

open-sse/executors/cliproxyapi.ts

@@ -408,12 +408,13 @@ export class CliproxyapiExecutor extends BaseExecutor {
 
     input.log?.info?.("CPA", `CLIProxyAPI → ${url} (model: ${input.model}, shape: ${shape})`);
 
-    // _toolNameMap is an in-memory channel to chatCore for response-side
-    // tool name restoration; never send it over the wire.
+    // _toolNameMap and _namespaceToolIdentityMap are in-memory channels to
+    // chatCore for response-side tool name restoration; never send them over
+    // the wire.
     const wireBody =
       transformedBody && typeof transformedBody === "object"
         ? JSON.stringify(transformedBody, (key, value) =>
-            key === "_toolNameMap" ? undefined : value
+            key === "_toolNameMap" || key === "_namespaceToolIdentityMap" ? undefined : value
           )
         : JSON.stringify(transformedBody);
 

open-sse/handlers/chatCore.ts

@@ -207,7 +207,6 @@ import { stageTrace } from "./chatCore/stageTrace.ts";
 import { attachCompressionUsageReceiptAfterAnalytics as attachCompressionUsageReceiptAfterAnalyticsFor } from "./chatCore/compressionUsageReceipt.ts";
 import { prepareUpstreamBody } from "./chatCore/upstreamBody.ts";
 import { getQuotaScopeLabelForProvider } from "../services/antigravityQuotaFamily.ts";
-
 import {
   getCallLogPipelineCaptureStreamChunks,
   getCallLogPipelineMaxSizeBytes,
@@ -367,9 +366,7 @@ import {
   isTpmExhausted,
   isRpmExhausted,
 } from "../services/geminiRateLimitTracker.ts";
-
 import { isSmallEnoughForSemanticCache } from "../utils/estimateSize.ts";
-
 /**
  * Core chat handler - shared between SSE and Worker
  * Returns { success, response, status, error } for caller to handle fallback
@@ -389,10 +386,8 @@ import { isSmallEnoughForSemanticCache } from "../utils/estimateSize.ts";
  * @param {boolean} options.isCombo - Whether this request is from a combo
  * @param {string} options.connectionId - Connection ID for settings lookup
  */
-
 // extractSystemRoleMessages extracted to chatCore/claudeSystemRole.ts (#3501); re-exported above so
 // existing importers (e.g. tests/unit/system-role-extraction.test.ts) keep resolving it from here.
-
 export async function handleChatCore({
   body,
   modelInfo,
@@ -428,7 +423,6 @@ export async function handleChatCore({
       /* fail open */
     }
   }
-
   // Per-request model-routing metadata (first extracted slice of the request-setup phase).
   const { apiFormat, customModelTargetFormat, requestedModel } = resolveChatCoreRequestSetup(
     modelInfo,
@@ -442,7 +436,6 @@ export async function handleChatCore({
   // (not Math.random) purely to satisfy CodeQL js/insecure-randomness — this id
   // is a log-correlation token, not a security secret.
   const traceId = globalThis.crypto.randomUUID().slice(0, 6);
-
   // Emit request.started event for real-time dashboard
   setImmediate(() => {
     emit("request.started", {
@@ -526,7 +519,6 @@ export async function handleChatCore({
       `long-running goal mode enabled: readinessMax=${agentGoalPolicy.readinessMaxTimeoutMs}ms streamRecovery=${agentGoalPolicy.streamRecoveryEnabled}`
     );
   }
-
   let effectiveServiceTier: EffectiveServiceTier = "standard";
   // Codex service-tier resolvers extracted to chatCore/serviceTier.ts (#3501); bind the per-request
   // provider/credentials once and delegate so the existing call sites stay byte-identical.
@@ -555,7 +547,6 @@ export async function handleChatCore({
       })
     ).catch(() => {});
   };
-
   // Key-health updater extracted to chatCore/keyHealth.ts (#3501); bind the per-request log once
   // and delegate so the existing call sites stay byte-identical.
   const recordKeyHealthStatus = (
@@ -563,11 +554,9 @@ export async function handleChatCore({
     creds: Record<string, unknown> | null | undefined,
     transport?: string
   ): void => recordKeyHealthStatusFor(status, creds, log, transport);
-
   const persistCodexQuotaState = async (headers: Record<string, string> | null, status = 0) => {
     const currentConnectionId = getCurrentConnectionId();
     if (provider !== "codex" || !currentConnectionId || !headers) return;
-
     try {
       const existingProviderData =
         credentials?.providerSpecificData && typeof credentials.providerSpecificData === "object"
@@ -582,28 +571,23 @@ export async function handleChatCore({
         status,
       });
       if (!built) return;
-
       if (built.exhaustionLog) {
         log?.debug?.("CODEX", built.exhaustionLog);
       }
-
       // Invalidate the preflight cache for this connection so the next
       // isModelAvailable check fetches fresh quota data.
       if (status === 429) {
         invalidateCodexQuotaCache(currentConnectionId);
       }
-
       await updateProviderConnection(currentConnectionId, {
         providerSpecificData: built.nextProviderData,
       });
-
       credentials.providerSpecificData = built.nextProviderData;
     } catch (err) {
       const errMessage = err instanceof Error ? err.message : String(err);
       log?.debug?.("CODEX", `Failed to persist codex quota state: ${errMessage}`);
     }
   };
-
   // ── Phase 9.2: Idempotency check ──
   // Resolve the idempotency key once here and reuse it at the Phase 9.2 save site below,
   // rather than re-deriving it. (#3821-review LEDGER-6)
@@ -622,13 +606,11 @@ export async function handleChatCore({
   if (idempotencyHit) {
     return idempotencyHit;
   }
-
   // T07: Inject connectionId into credentials so executors can rotate API keys
   // using providerSpecificData.extraApiKeys (API Key Round-Robin feature)
   if (connectionId && credentials && !credentials.connectionId) {
     credentials.connectionId = connectionId;
   }
-
   // Endpoint/format resolution extracted to chatCore/requestFormat.ts (#3501); pure derivation
   // from the inbound request, destructured so every downstream use stays byte-identical.
   const {
@@ -2264,8 +2246,19 @@ export async function handleChatCore({
   // the latter is a Kiro/Claude passthrough alias channel with string values,
   // while namespace identities carry `{namespace, name}` for the #7936 response
   // seam. Extract first because Kiro merge may reuse `_toolNameMap` below.
+  //
+  // #9780 — prefer the dedicated channel: on a pivot the openai->claude/gemini
+  // step publishes its own alias map on `_toolNameMap`, so that property alone
+  // yields aliases here. The `_toolNameMap` read stays as the fallback for the
+  // non-pivot producers (executors/base.ts, cliproxyapi.ts, antigravity).
+  const namespaceIdentityMap = translatedBody._namespaceToolIdentityMap;
   const requestToolIdentityMap =
-    translatedBody._toolNameMap instanceof Map ? translatedBody._toolNameMap : null;
+    namespaceIdentityMap instanceof Map
+      ? namespaceIdentityMap
+      : translatedBody._toolNameMap instanceof Map
+        ? translatedBody._toolNameMap
+        : null;
+  delete translatedBody._namespaceToolIdentityMap;
   delete translatedBody._toolNameMap;
 
   // Kiro: sanitize tool schemas before dispatch. Kiro returns 400 "Improperly
@@ -5025,7 +5018,6 @@ export async function handleChatCore({
     }),
   };
 }
-
 export function isTokenExpiringSoon(expiresAt, bufferMs = 5 * 60 * 1000) {
   if (!expiresAt) return false;
   const expiresAtMs = new Date(expiresAt).getTime();

open-sse/translator/index.ts

@@ -352,7 +352,27 @@ export function translateRequest(
                   ...(hasProvider ? { _provider: provider } : {}),
                 }
               : credentials;
-          result = fromOpenAI(model, result, stream, translationCredentials);
+          // #9780 — carry the Responses namespace identity map across the pivot.
+          // Target translators return a brand-new object (buildKiroPayload et
+          // al.), dropping the non-enumerable property step 1 attached; the
+          // #7936 seam then gets null and namespace sub-tool calls come back
+          // flattened, which Codex rejects with `unsupported call: <name>`.
+          const identityMap = (result as Record<string, unknown>)._namespaceToolIdentityMap;
+          const translated = fromOpenAI(model, result, stream, translationCredentials);
+          if (
+            identityMap instanceof Map &&
+            translated &&
+            typeof translated === "object" &&
+            !((translated as Record<string, unknown>)._namespaceToolIdentityMap instanceof Map)
+          ) {
+            Object.defineProperty(translated, "_namespaceToolIdentityMap", {
+              value: identityMap,
+              enumerable: false,
+              configurable: true,
+              writable: true,
+            });
+          }
+          result = translated;
         }
       }
     }

open-sse/translator/request/openai-responses.ts

@@ -752,8 +752,19 @@ export function openaiResponsesToOpenAIRequest(
   delete result.prompt_cache_retention;
 
   if (namespaceToolIdentityMap.size > 0) {
-    // chatCore extracts and deletes this transient side channel before dispatch.
+    // chatCore extracts and deletes these transient side channels before dispatch.
     // Non-enumerability keeps internal request metadata off the upstream wire.
+    //
+    // Two properties on purpose (#9780): `_toolNameMap` is also the alias
+    // channel for openai-to-claude/gemini, which overwrite it on a pivot, so
+    // the identity map needs a name of its own. `_toolNameMap` stays populated
+    // for the existing consumers (executors/base.ts, cliproxyapi, antigravity).
+    Object.defineProperty(result, "_namespaceToolIdentityMap", {
+      value: namespaceToolIdentityMap,
+      enumerable: false,
+      configurable: true,
+      writable: true,
+    });
     Object.defineProperty(result, "_toolNameMap", {
       value: namespaceToolIdentityMap,
       enumerable: false,

tests/unit/9780-namespace-identity-pivot.test.ts

@@ -0,0 +1,155 @@
+// #9780 — the Responses namespace identity map must survive the hub-and-spoke
+// pivot in translator/index.ts. Step 1 flattens namespace sub-tools (#8295) and
+// records `{namespace, name}`; step 2 returns a new object and used to drop it,
+// leaving the #7936 seam with null and Codex rejecting `unsupported call`.
+// A naive copy-through is not an option: openai-to-claude/gemini publish their
+// own alias map on `_toolNameMap`, hence the dedicated channel asserted here.
+import test from "node:test";
+import assert from "node:assert/strict";
+
+await import("../../open-sse/translator/bootstrap.ts");
+const { translateRequest, initState } = await import("../../open-sse/translator/index.ts");
+const { openaiToOpenAIResponsesResponse } = await import(
+  "../../open-sse/translator/response/openai-responses.ts"
+);
+const { FORMATS } = await import("../../open-sse/translator/formats.ts");
+
+type NamespaceIdentity = { namespace: string; name: string };
+
+const NAMESPACE_REQUEST = {
+  model: "any-model",
+  instructions: "coding agent",
+  input: [{ type: "message", role: "user", content: [{ type: "input_text", text: "go" }] }],
+  tools: [
+    {
+      type: "namespace",
+      name: "functions",
+      tools: [
+        {
+          name: "exec",
+          description: "Run a shell command",
+          parameters: {
+            type: "object",
+            properties: { cmd: { type: "string" } },
+            required: ["cmd"],
+          },
+        },
+      ],
+    },
+  ],
+};
+
+function pivot(targetFormat: string): Record<string, unknown> {
+  return translateRequest(
+    "openai-responses",
+    targetFormat,
+    "any-model",
+    structuredClone(NAMESPACE_REQUEST),
+    true,
+    null,
+    null,
+    null
+  ) as Record<string, unknown>;
+}
+
+function identityOf(body: Record<string, unknown>) {
+  const map = body._namespaceToolIdentityMap;
+  assert.ok(map instanceof Map, "expected a _namespaceToolIdentityMap after the pivot");
+  return map as Map<string, NamespaceIdentity>;
+}
+
+test("#9780: namespace identity survives the openai-responses -> kiro pivot", () => {
+  const identity = identityOf(pivot("kiro"));
+
+  assert.equal(identity.size, 1);
+  assert.deepEqual(identity.get("functions__exec"), { namespace: "functions", name: "exec" });
+});
+
+test("#9780: namespace identity survives the openai-responses -> cursor pivot", () => {
+  const identity = identityOf(pivot("cursor"));
+
+  assert.deepEqual(identity.get("functions__exec"), { namespace: "functions", name: "exec" });
+});
+
+// Regression guard: these two appeared to "keep" a map before the fix, but it
+// was the alias map.
+for (const target of ["claude", "gemini"]) {
+  test(`#9780: ${target} pivot keeps its alias map AND the namespace identity`, () => {
+    const body = pivot(target);
+    const identity = identityOf(body);
+
+    assert.deepEqual(identity.get("functions__exec"), { namespace: "functions", name: "exec" });
+
+    // The alias channel must be untouched: string values, not identities.
+    const aliases = body._toolNameMap;
+    assert.ok(aliases instanceof Map, `${target} must still publish its alias map`);
+    for (const value of (aliases as Map<string, unknown>).values()) {
+      assert.equal(typeof value, "string", `${target} alias values must stay strings`);
+    }
+  });
+}
+
+// Same-format requests are never flattened, so an absent map is correct here.
+test("#9780: same-format openai-responses request is not flattened at all", () => {
+  const body = pivot("openai-responses");
+  const tools = body.tools as Array<Record<string, unknown>>;
+
+  assert.equal(tools[0].type, "namespace");
+  assert.equal((tools[0].tools as Array<{ name: string }>)[0].name, "exec");
+  assert.equal(body._namespaceToolIdentityMap, undefined);
+});
+
+test("#9780: the identity channel is non-enumerable and never serializes", () => {
+  const body = pivot("kiro");
+
+  assert.ok(body._namespaceToolIdentityMap instanceof Map);
+  assert.equal(
+    Object.prototype.propertyIsEnumerable.call(body, "_namespaceToolIdentityMap"),
+    false
+  );
+  assert.equal("_namespaceToolIdentityMap" in JSON.parse(JSON.stringify(body)), false);
+});
+
+// End-to-end: request pivot + response seam, i.e. what the Codex adjudicator
+// actually receives. Before the fix every target emitted `functions__exec` with
+// no namespace, which is the reported `unsupported call`.
+for (const target of ["kiro", "cursor", "claude", "gemini"]) {
+  test(`#9780: ${target} round-trip returns the declared name and its namespace`, () => {
+    const body = pivot(target);
+    const state = initState(FORMATS.OPENAI_RESPONSES) as Record<string, unknown>;
+    state.requestToolIdentityMap = body._namespaceToolIdentityMap;
+
+    // The upstream echoes the flattened wire name (#8295).
+    const events = openaiToOpenAIResponsesResponse(
+      {
+        id: "chatcmpl-9780",
+        model: "any-model",
+        choices: [
+          {
+            index: 0,
+            delta: {
+              tool_calls: [
+                {
+                  index: 0,
+                  id: "call_9780",
+                  type: "function",
+                  function: { name: "functions__exec", arguments: '{"cmd":"git status"}' },
+                },
+              ],
+            },
+            finish_reason: "tool_calls",
+          },
+        ],
+        usage: { prompt_tokens: 1, completion_tokens: 1, total_tokens: 2 },
+      },
+      state
+    ) as Array<{ event: string; data: { item?: NamespaceIdentity } }>;
+
+    const added = events.find((e) => e.event === "response.output_item.added")?.data.item;
+    assert.ok(added, "expected response.output_item.added");
+    assert.deepEqual(
+      { name: added.name, namespace: added.namespace },
+      { name: "exec", namespace: "functions" }
+    );
+  });
+}