Merge remote-tracking branch 'origin/main' into cl-11727-restore-the-live-codex-model-catalog-fetch-for-chatgpt

The comment above CODEX_DEFAULT_MODELS conflicted with #1580; this keeps both:
the live-catalog fallback role and the Plus-catalog provenance. The list is
unchanged.

Author: bwachman

.corbits-triage.yml

@@ -0,0 +1,141 @@
+# Corbits Triage policy.
+rules:
+  - check: max-files-changed
+    text: Flag pull requests over 30 files changed.
+    max: 30
+    onFail: label
+    reason: A large change is a risk signal, not a hold.
+    evidence:
+      - maintainer-stated
+  - check: max-lines-changed
+    text: Flag pull requests over 5,000 lines changed.
+    max: 5000
+    onFail: label
+    reason: A large change is a risk signal, not a hold.
+    evidence:
+      - maintainer-stated
+  - check: no-merge-conflicts
+    text: Flag merge conflicts with the base branch.
+    reason: A pull request with conflicts cannot be merged as it is.
+    evidence:
+      - maintainer-stated
+  - check: ci-green
+    text: CI must pass before merge.
+    reason: A change is not ready until CI passes.
+    evidence:
+      - maintainer-stated
+  - check: conventional-commits
+    text: Titles and commits follow Conventional Commits.
+    reason: Titles and commits use the conventional form.
+    evidence:
+      - maintainer-stated
+  - check: hidden-content
+    text: Flag text hidden in the pull request.
+    reason: Hidden text can steer a review.
+    evidence:
+      - maintainer-stated
+  - check: no-duplicate
+    text: Flag a pull request whose diff or commits match another open one.
+    reason: Two pull requests for the same change waste a review.
+    evidence:
+      - maintainer-stated
+  - check: sensitive-paths
+    text: Route schema, auth, CI and lockfile changes to a person.
+    reason: Those changes reach beyond what the diff shows.
+    evidence:
+      - maintainer-stated
+  - check: no-forbidden-paths
+    text: Never accept changes to node_modules/**.
+    paths:
+      - node_modules/**
+    reason: Dependencies are not edited by hand.
+    evidence:
+      - maintainer-stated
+  - id: no-ai-attribution
+    check: text
+    text: Titles, descriptions and commits carry no AI attribution.
+    fields: [title, description, commits]
+    failWhen: contains
+    phrases:
+      - "Co-Authored-By:"
+      - "Claude-Session:"
+      - "claude.ai/code/session"
+      - "noreply@anthropic.com"
+      - "Generated with Claude"
+      - "Generated with [Claude"
+    onFail: label
+    reason: No AI co-author lines, session trailers, or generated-with footers.
+    evidence:
+      - maintainer-stated
+  - id: conventions
+    check: code-patterns
+    text: Added code follows the repository's conventions.
+    presets:
+      - convention
+    onFail: label
+    reason: No invented fallbacks, nested ternaries, immediately invoked functions, promise chains, empty catches, or suppression comments.
+    evidence:
+      - maintainer-stated
+  - id: no-process-env-in-tests
+    check: code-patterns
+    text: Tests do not read process.env.
+    onFail: label
+    reason: A test must not depend on the environment.
+    evidence:
+      - maintainer-stated
+    patterns:
+      - id: no-process-env-in-tests
+        language: typescript
+        message: Reads process.env in a test.
+        files:
+          [
+            "**/*.test.ts",
+            "**/*.test.tsx",
+            "**/*.test.js",
+            "**/*.test.jsx",
+            "**/*.test.mjs",
+            "**/*.test.cjs",
+          ]
+        rule:
+          pattern: process.env
+      - id: no-process-env-in-tests
+        language: javascript
+        message: Reads process.env in a test.
+        files:
+          [
+            "**/*.test.ts",
+            "**/*.test.tsx",
+            "**/*.test.js",
+            "**/*.test.jsx",
+            "**/*.test.mjs",
+            "**/*.test.cjs",
+          ]
+        rule:
+          pattern: process.env
+actions:
+  pass:
+    - do: approve
+      when:
+        author:
+          - users:
+              - thegreataxios
+              - brianjfox
+    - do: request-review
+      target:
+        to: users
+        users:
+          - thegreataxios
+      when:
+        author:
+          - external
+  fail:
+    - do: request-changes
+  unsure:
+    - do: assign
+      target:
+        to: users
+        users:
+          - thegreataxios
+      when:
+        author:
+          - maintainers

CHANGELOG.md

@@ -20,6 +20,20 @@ release that deserves narration. Passing `--notes <file>` overrides both for
 the GitHub body. Sections below this line predate generation and were written
 by hand.
 
+## [0.3.48] - 2026-10-30
+
+### What's Changed
+* fix(agent): match tool_search queries against tool parameters by @bwachman in https://github.com/corbitsdev/corbits-code/pull/1573
+* fix(agent): match tool names across a trailing plural s by @bwachman in https://github.com/corbitsdev/corbits-code/pull/1574
+* fix(reactor): keep rejected and timed-out approval results across a rebuild by @bwachman in https://github.com/corbitsdev/corbits-code/pull/1575
+* fix(session): settle a dropped approval on the rebuilt agent without a new turn by @bwachman in https://github.com/corbitsdev/corbits-code/pull/1576
+* chore(deps): bump @corbits/oauth-core to 0.3.3 (CL-11513) by @TheGreatAxios in https://github.com/corbitsdev/corbits-code/pull/1572
+* fix(mcp): keep padded optional arguments from filtering MCP reads by @bwachman in https://github.com/corbitsdev/corbits-code/pull/1578
+* feat(review): post one plain-language review by @TheGreatAxios in https://github.com/corbitsdev/corbits-code/pull/1582
+
+
+**Full Changelog**: https://github.com/corbitsdev/corbits-code/compare/v0.3.47...v0.3.48
+
 ## [0.3.47] - 2026-10-26
 
 ### What's Changed

bun.lock

@@ -371,7 +371,7 @@ describe("integration — late MCP dispatch", () => {
   );
 
   test.serial(
-    "padded optional MCP arguments reach the server minus the values its schema rejects",
+    "padded optional arguments to a read MCP tool reach the server without rejected placeholders or null",
     async () => {
       const received: Record<string, unknown>[] = [];
       await withSession(async (session) => {
@@ -428,7 +428,6 @@ describe("integration — late MCP dispatch", () => {
           team: "",
           fields: [],
           priority: 0,
-          assignee: null,
           release: "v0345-ef53e73667fa",
         },
       ]);

e2e/mcp-late-dispatch.test.ts

@@ -1,6 +1,6 @@
 {
   "name": "@corbits/code",
-  "version": "0.3.47",
+  "version": "0.3.48",
   "description": "Single-process coding agent CLI built on the Interchange runtime (Bun + TypeScript)",
   "private": true,
   "license": "SEE LICENSE IN LICENSE.md",
@@ -90,7 +90,7 @@
     "@anthropic-ai/sandbox-runtime": "0.0.79",
     "@corbits/codex-provider": "0.1.3",
     "@corbits/meta-provider": "0.1.1",
-    "@corbits/oauth-core": "0.3.2",
+    "@corbits/oauth-core": "0.3.3",
     "@corbits/openai-responses": "0.2.2",
     "@corbits/xai-provider": "0.1.4",
     "@intx/agent": "workspace:*",

package.json

@@ -52,29 +52,36 @@ Deeper work (file-by-file behavior, architecture, coherence) can go to sub-agent
 
 Raise what helps the author. Flag real defects and architecture that would constrain future work. Skip hypothetical problems in cases the code need not handle, stylistic nits, and vague remarks like "could be cleaner".
 
-- Every finding has a `path:line` and a concrete failure mode. Severity is the review action, not adjectives.
+- Every finding is one of **blocking** (breaks the contract or the behavior), **should-fix** (a real defect that should change in this diff), or **informational** (true, but not something this diff has to change; the same rank as file-for-later). Severity is the review action, not an adjective.
+- Cite a line range, `path:start-end`, not a single line unless the failure is one line. A range is the lines a reader opens to see the failure.
 - Describe the branch as it stands now, in present tense, from `git diff <base>...HEAD`, not the journey or the PR description.
-- Write to a colleague: propose rather than command, say why, describe consequences instead of "wrong" or "broken", and suggest the one-liner when the fix is small.
+- Write the description for a person: what goes wrong, and the consequence. No finder jargon, no pass names.
 - **Cite the check.** Every affirmative claim ("tests pass", "no regressions") names the command or file range that proved it. Otherwise run the check, strike the claim, or narrow it to what you examined ("read `lock.go:42-68`, no acquire-while-holding cycle there").
 
 ## 6. Report and post
 
-Summarize with the findings. When the target has an open GitHub PR, post the finished review on it; a review that only lives in chat is not done. Post when the operator asked for a PR review, when `/implement`'s whole-branch gate runs on a branch with an open PR, or when a record on the PR is the purpose. Do not post a private local read, and do not reopen the analysis while writing the body.
+Summarize with the findings. When the target has an open GitHub PR, post the finished review on it; a review that only lives in chat is not done. Post when the operator asked for a PR review, when `/implement`'s whole-branch gate runs on a branch with an open PR, or when a record on the PR is the purpose. Do not post a private local read. Do not implement fixes. Do not reopen the analysis while writing the body.
+
+Wait until every pass Dispatch started has reported. Passes do not post to GitHub and do not approve, comment, or request changes.
+
+Then post exactly one review with `gh pr review`. One verdict for the whole diff. Use one of `--approve`, `--comment`, or `--request-changes`, and swap that flag to match the verdict:
 
 ```bash
 gh pr review <number-or-url> --comment --body-file <file>
 ```
 
-Use `--approve` or `--request-changes` for the verdict. Shape:
-
-```markdown
-## Review · <Approve | Comment | Request changes>
+The body is one short paragraph of what the branch does now, then a bullet list. Nothing else. Do not name the passes. Do not use finder jargon.
 
-<one line: what the branch does>
+Each bullet is the severity, then the description, then the line range:
 
-### Findings
+```markdown
+<what the branch does now, in plain language>
 
-- `path/to/file.ts:12` — concrete problem and why it matters
+- **blocking** — <what goes wrong> `path/to/file.ts:12-18`
+- **should-fix** — <what goes wrong> `path/to/file.ts:40-44`
+- **informational** — <what is true, and why this diff need not change it> `path/to/file.ts:7`
 ```
 
-A clean review says "No findings." after the one line. No throat-clearing, praise to soften a finding, emoji, or "nit:" taste. When more than one lens produced a distinct judgment (reviewer, planner waiver, warden), each posts its own labelled review, and only the lens that owns the merge verdict may approve. Paste the review URL back to the operator. Posting does not change Linear state. Remove the PR worktree afterward.
+Request changes when any bullet is blocking or should-fix. Comment when every bullet is informational. Approve when there are no bullets, and say "No findings." after the paragraph. No throat-clearing, praise, emoji, or "nit:".
+
+The note back to the operator is that same body plus the review URL. Do not write a second, more technical summary in chat. Posting does not change Linear state. Remove the PR worktree afterward.

plugins/corbits-skills/skills/review/SKILL.md

@@ -163,6 +163,78 @@ describe("normalizeToolDefinitionsForProvider", () => {
     }
   });
 
+  describe("optional MCP properties", () => {
+    const listIssues: ToolDefinition = {
+      name: "mcp__linear__list_issues",
+      description: "[linear] list issues",
+      inputSchema: {
+        type: "object",
+        properties: {
+          release: { type: "string" },
+          priority: { type: "number" },
+        },
+        required: ["release"],
+      },
+    };
+    const builtin: ToolDefinition = {
+      ...listIssues,
+      name: "list_issues",
+    };
+
+    test.each([
+      { providerName: "codex/plus", model: "gpt-5.6-terra" },
+      { providerName: "opencode-go", model: "gpt-5.1" },
+      { providerName: "codex", model: "codex-mini-latest" },
+    ])("$providerName $model advertises null on MCP tools only", (ctx) => {
+      const [mcp, own] = normalizeToolDefinitionsForProvider(
+        [listIssues, builtin],
+        ctx,
+      );
+      expect(mcp?.inputSchema).toEqual({
+        type: "object",
+        properties: {
+          release: { type: "string" },
+          priority: { type: ["number", "null"] },
+        },
+        required: ["release"],
+      });
+      expect(own).toBe(builtin);
+    });
+
+    test.each([
+      { providerName: "anthropic", model: "claude-sonnet-5-5" },
+      { providerName: "moonshot", model: "kimi-k2" },
+    ])("$providerName leaves MCP tools as they are", (ctx) => {
+      const defs = [listIssues];
+      expect(normalizeToolDefinitionsForProvider(defs, ctx)).toBe(defs);
+    });
+
+    test("grok sanitizes MCP tools without advertising null", () => {
+      const [out] = normalizeToolDefinitionsForProvider([listIssues], {
+        providerName: "xai/default-2",
+        model: "grok-4.6",
+      });
+      expect(out?.inputSchema).toEqual(listIssues.inputSchema);
+    });
+
+    test("returns the same array when no MCP property changes", () => {
+      const nullable: ToolDefinition = {
+        ...listIssues,
+        inputSchema: {
+          type: "object",
+          properties: { assignee: { type: ["string", "null"] } },
+        },
+      };
+      const defs = [nullable, builtin];
+      expect(
+        normalizeToolDefinitionsForProvider(defs, {
+          providerName: "openai",
+          model: "gpt-5.6",
+        }),
+      ).toBe(defs);
+    });
+  });
+
   test("grok sanitizes untyped properties and strips $schema", () => {
     const untyped: ToolDefinition = {
       name: "submit_result",