Merge remote-tracking branch 'origin/main' into HEAD

# Conflicts:
#	packages/sdk-typescript/scripts/build.js

Author: yiliang114

.github/workflows/qwen-code-pr-review.yml

@@ -159,9 +159,17 @@ jobs:
           fi
 
   review-config:
+    # Bot-requested review_requested only: a CODEOWNERS-covered PR open
+    # auto-requests every owner individually (#8945), spawning one
+    # review_requested run per owner. Only the run where the bot itself is
+    # the requested reviewer can reach review-pr, so the human-requested
+    # siblings must skip here instead of each spending a runner. KEEP IN
+    # SYNC with the review_requested clauses in precheck-pr.if and
+    # authorize.if, and with the bot_login constant below.
     if: |-
       github.event_name == 'pull_request_target' &&
-      github.event.action == 'review_requested'
+      github.event.action == 'review_requested' &&
+      github.event.requested_reviewer.login == 'qwen-code-ci-bot'
     runs-on: '${{ (github.repository == ''QwenLM/qwen-code'' && vars.MAINTAINER_ECS_RUNNER_DISABLED != ''true'') && fromJSON(''["self-hosted", "linux", "x64", "ecs-qwen"]'') || fromJSON(''["ubuntu-latest"]'') }}'
     permissions: {}
     outputs:
@@ -225,13 +233,21 @@ jobs:
     # Only run for PR-target events and supported command comments — not every
     # unrelated comment — to avoid spawning a job per comment. The downstream
     # `if`s still do the exact command body match; this prefix is just a filter.
+    # review_requested must additionally request the bot itself: a
+    # CODEOWNERS-covered PR open auto-requests every owner individually
+    # (#8945), and precheck-pr's identical predicate only covers fork PRs —
+    # without this clause each same-repo sibling run spends an authorize job
+    # (permission API + runner slot) before review-pr no-op exits.
     if: |-
       !cancelled() &&
       (github.event_name != 'pull_request_target' ||
        github.event.action != 'closed') &&
       (github.event_name != 'pull_request_target' ||
        github.event.pull_request.head.repo.full_name == github.repository ||
        needs.precheck-pr.outputs.decision == 'allow_triage') &&
+      (github.event_name != 'pull_request_target' ||
+       github.event.action != 'review_requested' ||
+       github.event.requested_reviewer.login == 'qwen-code-ci-bot') &&
       (github.event_name == 'pull_request_target' ||
        (github.event_name == 'workflow_dispatch' &&
         github.event.inputs.command == 'resolve') ||

docs/design/2026-08-13-review-platform-provider-abstraction.md

@@ -124,16 +124,16 @@ These events are workspace-keyed, not session-keyed. The session reducer treats
 
 ### Turn lifecycle / assistant pushes
 
-| Type                  | Direction | Trigger                                                                                                             | Key payload fields                                                                                                                                                                               |
-| --------------------- | --------- | ------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
-| `prompt_cancelled`    | S->C      | Prompt was cancelled through explicit `cancelSession` route **or** originator SSE disconnect                        | Envelope stamps `originatorClientId` for the canceling client. This means "cancellation requested", not "cancellation confirmed". Peer subscribers learn that the prompt has ended.              |
-| `turn_complete`       | S->C      | A turn completed successfully                                                                                       | `sessionId, stopReason, promptId?`. `promptId` links to non-blocking prompt responses (`202`). The SDK matches SSE events to the originating prompt through it.                                  |
-| `turn_error`          | S->C      | A turn failed                                                                                                       | `sessionId, message, code?, promptId?`; same `promptId` correlation mechanism.                                                                                                                   |
-| `session_rewound`     | S->C      | `POST /session/:id/rewind` succeeded                                                                                | `sessionId, promptId, targetTurnIndex, filesChanged[], filesFailed[], originatorClientId?`                                                                                                       |
-| `session_branched`    | S->C      | `POST /session/:id/branch` created a branch from an existing session                                                | `sourceSessionId, newSessionId, displayName, originatorClientId?`                                                                                                                                |
-| `followup_suggestion` | S->C      | ACP child generated ghost-text follow-up suggestions after `end_turn`, forwarded over per-session SSE               | `sessionId, suggestion, promptId`; wire only carries suggestions whose `getFilterReason()===null`. Clients render them as input-placeholder ghost text and invalidate them on next `sendPrompt`. |
-| `user_shell_command`  | S->C      | User started a shell command through `POST /session/:id/shell`; fanned out to other subscribers in the same session | `sessionId, command, shellId, originatorClientId?`. There is no typed `DaemonXxxData` interface yet; `asKnownDaemonEvent` returns `undefined` and the UI normalizer parses it ad hoc.            |
-| `user_shell_result`   | S->C      | Result of the shell command above                                                                                   | `sessionId, shellId, exitCode, output, aborted`. Same ad hoc parsing note as `user_shell_command`.                                                                                               |
+| Type                  | Direction | Trigger                                                                                                             | Key payload fields                                                                                                                                                                                    |
+| --------------------- | --------- | ------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `prompt_cancelled`    | S->C      | Prompt was cancelled through explicit `cancelSession` route **or** originator SSE disconnect                        | Envelope stamps `originatorClientId` for the canceling client. This means "cancellation requested", not "cancellation confirmed". Peer subscribers learn that the prompt has ended.                   |
+| `turn_complete`       | S->C      | A turn completed successfully                                                                                       | `sessionId, stopReason, promptId?, branchPoint?`. `promptId` links to non-blocking prompt responses (`202`). Eligible completed turns include `branchPoint: { assistantRecordUuid, checkpointUuid }`. |
+| `turn_error`          | S->C      | A turn failed                                                                                                       | `sessionId, message, code?, promptId?`; same `promptId` correlation mechanism.                                                                                                                        |
+| `session_rewound`     | S->C      | `POST /session/:id/rewind` succeeded                                                                                | `sessionId, promptId, targetTurnIndex, filesChanged[], filesFailed[], originatorClientId?`                                                                                                            |
+| `session_branched`    | S->C      | Legacy compatibility event; the current branch endpoint returns its result directly and does not publish this event | `sourceSessionId, newSessionId, displayName, originatorClientId?`. Readers retain support for older producers.                                                                                        |
+| `followup_suggestion` | S->C      | ACP child generated ghost-text follow-up suggestions after `end_turn`, forwarded over per-session SSE               | `sessionId, suggestion, promptId`; wire only carries suggestions whose `getFilterReason()===null`. Clients render them as input-placeholder ghost text and invalidate them on next `sendPrompt`.      |
+| `user_shell_command`  | S->C      | User started a shell command through `POST /session/:id/shell`; fanned out to other subscribers in the same session | `sessionId, command, shellId, originatorClientId?`. There is no typed `DaemonXxxData` interface yet; `asKnownDaemonEvent` returns `undefined` and the UI normalizer parses it ad hoc.                 |
+| `user_shell_result`   | S->C      | Result of the shell command above                                                                                   | `sessionId, shellId, exitCode, output, aborted`. Same ad hoc parsing note as `user_shell_command`.                                                                                                    |
 
 ## Architecture
 

docs/design/web-shell/assistant-response-session-branching.md

@@ -299,9 +299,9 @@ For PR reviews the manifest is read from the merge base, so the PR under review
 
 ## Issue Fidelity
 
-For bugfix PRs, the Issue Fidelity agent fetches issue evidence directly instead of relying on PR description text. It uses `gh pr view <pr> --repo <owner/repo> --json closingIssuesReferences` for GitHub's strong closing-issue metadata, then `gh issue view <number> --repo <issue_owner>/<issue_repo> --json title,body,comments` for the original report and discussion — the `--json` form includes the issue **body** (the reporter's original repro), which `--comments` alone omits, and the issue's own repository is read from each reference (a PR can close an issue in a different repo). This agent runs only for PR targets; local-diff and file-path reviews skip it.
+For bugfix PRs, the Issue Fidelity agent fetches issue evidence directly instead of relying on PR description text. It runs the `qwen review issue-context <pr> --repo <owner/repo> --out <file>` subcommand, which resolves GitHub's strong closing-issue metadata and then fetches each referenced issue's title, **body** (the reporter's original repro), and full comment thread — each from the issue's own repository (a PR can close an issue in a different repo). This agent runs only for PR targets; local-diff and file-path reviews skip it.
 
-`closingIssuesReferences` is a discovery hint rather than proof the author linked the right issue: if it is empty but the PR references an apparent target issue, the agent still fetches it after judging relevance. Fetched issue text is treated as untrusted data (facts extracted, embedded instructions ignored). For relevant issues, the original reproduction, observed payload, expected behavior, and maintainer comments are treated as the highest-priority evidence for whether the PR fixes the right problem.
+The closing-issue set is a discovery hint rather than proof the author linked the right issue: if it is empty but the PR references an apparent target issue, the agent still fetches it after judging relevance (re-running with `--issue <n>`; a bare number resolves in the PR's repo, while `--issue <owner>/<repo>#<n>` fetches a cross-repo reference from its own repo). Fetched issue text is treated as untrusted data (facts extracted, embedded instructions ignored). For relevant issues, the original reproduction, observed payload, expected behavior, and maintainer comments are treated as the highest-priority evidence for whether the PR fixes the right problem.
 
 If the issue evidence shows an upstream service or provider returned malformed data outside the client contract, client-side parser or sanitizer changes are not treated as a valid root-cause fix unless a maintainer explicitly requested a defensive workaround. A test that replays malformed upstream output proves only that the workaround handles that shape; it does not prove the workaround is architecturally appropriate.
 
@@ -361,7 +361,7 @@ Medium- and high-effort reviews also save a structured JSON companion with the s
 
 The deterministic halves of the pipeline — argument parsing (`qwen review parse-args`) and the event/body decision (`qwen review compose-review`) — are tested subcommands rather than prompt text, so `--effort` grammar, `--comment` forcing, verdict caps, and downgrade behavior are pinned by unit tests and cannot drift with the model.
 
-**GitHub Enterprise:** reviewing a PR URL on a non-`github.com` host routes every GitHub call at that host — the review subcommands (`fetch-pr`, `pr-context`, `comment-status`, `presubmit`) accept `--host` and set it in code, so a forgotten host cannot silently retarget the review at `github.com`.
+**GitHub Enterprise:** reviewing a PR URL on a non-`github.com` host routes every GitHub call at that host — the review subcommands (`match-remote`, `meta`, `fetch-pr`, `pr-context`, `comment-status`, `issue-context`, `fetch-diff`, `comment-body`, `plan-diff`, `test-plan`, `presubmit`, `compose-review`, `submit`, `publish-assets`) accept `--host` and set it in code, so a forgotten host cannot silently retarget the review at `github.com`.
 
 Every run ends with one machine-readable line (`Review complete: <target> — <disposition>`), so scripts and CI wrappers can detect completion and outcome with a single `^Review complete: ` match.
 

docs/developers/daemon/09-event-schema.md

@@ -955,6 +955,78 @@ describePOSIX('qwen serve — Last-Event-ID resume', () => {
   }, 60_000);
 });
 
+describePOSIX('qwen serve — historical Assistant response branch', () => {
+  it('creates, opens, and continues a branch through the real daemon', async () => {
+    const source = await client.createOrAttachSession({
+      workspaceCwd: workspaceDir,
+      sessionScope: 'thread',
+    });
+    const first = await client.prompt(source.sessionId, {
+      prompt: [{ type: 'text', text: 'historical branch turn one' }],
+    });
+    expect(first.branchPoint).toBeDefined();
+    if (!first.branchPoint) return;
+
+    await client.prompt(source.sessionId, {
+      prompt: [{ type: 'text', text: 'historical branch turn two' }],
+    });
+    await client.prompt(source.sessionId, {
+      prompt: [{ type: 'text', text: 'historical branch turn three' }],
+    });
+
+    const branched = await client.branchSession(source.sessionId, {
+      atRecordId: first.branchPoint.checkpointUuid,
+    });
+    const branchBeforeContinue = await client.getSessionTranscriptPage(
+      branched.sessionId,
+      { limit: 500 },
+    );
+    const branchBeforeText = JSON.stringify(branchBeforeContinue.events);
+    expect(branchBeforeText).toContain('historical branch turn one');
+    expect(branchBeforeText).not.toContain('historical branch turn two');
+    expect(branchBeforeText).not.toContain('historical branch turn three');
+
+    const sourceAfterBranch = await client.getSessionTranscriptPage(
+      source.sessionId,
+      { limit: 500 },
+    );
+    const sourceText = JSON.stringify(sourceAfterBranch.events);
+    expect(sourceText).toContain('historical branch turn one');
+    expect(sourceText).toContain('historical branch turn two');
+    expect(sourceText).toContain('historical branch turn three');
+
+    const loadedBranch = await client.loadSession(branched.sessionId);
+    await client.prompt(
+      branched.sessionId,
+      {
+        prompt: [{ type: 'text', text: 'continue the historical branch' }],
+      },
+      undefined,
+      loadedBranch.clientId,
+    );
+    const branchAfterContinue = await client.getSessionTranscriptPage(
+      branched.sessionId,
+      { limit: 500 },
+    );
+    expect(JSON.stringify(branchAfterContinue.events)).toContain(
+      'continue the historical branch',
+    );
+
+    // The source session must stay untouched by the fork's continuation.
+    const sourceAfterContinue = await client.getSessionTranscriptPage(
+      source.sessionId,
+      { limit: 500 },
+    );
+    const sourceAfterContinueText = JSON.stringify(sourceAfterContinue.events);
+    expect(sourceAfterContinueText).not.toContain(
+      'continue the historical branch',
+    );
+    expect(sourceAfterContinueText).toContain('historical branch turn one');
+    expect(sourceAfterContinueText).toContain('historical branch turn two');
+    expect(sourceAfterContinueText).toContain('historical branch turn three');
+  }, 90_000);
+});
+
 describePOSIX('qwen serve — daemon Todo Stop Guard replay', () => {
   it('continues after prompt admission without an SSE client and replays the bounded attempts', async () => {
     const session = await client.createOrAttachSession({