feat(autofix): auto-manage the bot's own fork PRs without a label - #7243
Merged
Conversation
A fork PR the autofix bot itself opened (its codex flow pushes to qwen-code-dev-bot/qwen-code) is the bot's own generated work — same author, same code provenance, and the bot holds write+ — so it is trust-equal to an in-repo bot PR. Requiring a manual autofix/takeover label on it was redundant: in-repo bot PRs are auto-managed with no label, and the takeover label exists to authorize EXTERNAL (human) fork authors, not the bot's own. Now a fork authored by AUTOFIX_BOT with 'Allow edits from maintainers' is admitted and managed WITHOUT a label: - Scan: fork candidates are unioned from bot-prs.json (the bot's own forks — --author AUTOFIX_BOT, so no label needed) AND the takeover-labeled list (non-bot forks, explicit opt-in). Both still require allow-edits and pass the per-candidate live write+ gate. - Eligibility: the fork chain no longer demands the takeover label when the author is the bot (the author check already exempts it); it still demands allow-edits + a live write+ author + a matching live head repo. autofix/skip still opts any such PR out. Non-bot forks are unchanged — they still need the explicit label. Tests: the fork-candidate union admits a bot fork (no label) + a labeled human fork, dropping no-allow-edits/in-repo/skip; the eligibility replay makes a bot fork with allow-edits eligible without a label and discards it without allow-edits. 62/62 + 12/12.
Collaborator
|
Thanks for the PR! Template looks good ✓ (minor: "Why it's needed" is covered under "What this PR does" + "Why it's safe" rather than its own heading — substance is all there, no action needed). Problem: observed and well-evidenced. #7208 and #7220 are concrete bot codex-flow fork PRs that sat unprocessed until someone manually applied Direction: aligned. The bot's own fork PRs are trust-equal to in-repo bot PRs (same author, same code provenance, bot holds Size: not applicable (no core module paths touched). Approach: the scope feels right — two surgical changes (union the bot's forks into the scan candidate pool, relax the eligibility fork-chain for bot-authored forks) plus matching test updates. No scope creep, no drive-by refactors. The Moving on to code review. 🔍 中文说明感谢贡献！ 模板完整 ✓（小瑕疵："Why it's needed" 的内容分散在 "What this PR does" 和 "Why it's safe" 中，但实质内容齐全，无需修改）。 问题：已观测且有据可查。#7208 和 #7220 是具体的 bot codex 流程 fork PR，因缺少 方向：对齐。bot 自己的 fork PR 与 in-repo bot PR 同等信任（同作者、同代码来源、bot 持有 规模：不适用（未触及核心模块路径）。 方案：范围合理——两处精准修改（将 bot fork 并入扫描候选池、放宽 bot 作者 fork 的资格链）加上对应的测试更新。无范围蔓延，无顺手重构。jq 并集中的 进入代码审查 🔍 — Qwen Code · qwen3.7-max Reviewed at |
Collaborator
Code ReviewIndependent proposal: I'd do exactly what this PR does — union Comparison with the diff: the PR matches this approach precisely. Verified the full eligibility chain at address time:
No correctness bugs, security holes, or regressions found. No new abstractions or utilities — the change extends existing jq and bash patterns in place. Comments updated to match. TestingThis is a GitHub Actions workflow change — there's no CLI behavior to drive in tmux. The behavioral replay tests are the real-scenario verification here: they extract the exact jq filters and bash eligibility logic verbatim from the workflow YAML and replay them against fixtures. Key behavioral coverage in the 62 tests:
中文说明代码审查独立方案： 我会做和这个 PR 完全一样的事——将 与 diff 对比： PR 完全匹配此方案。验证了 address 时的完整资格链：
无正确性 bug、安全漏洞或回归。无新抽象或工具——变更就地扩展现有 jq 和 bash 模式。注释已同步更新。 测试这是 GitHub Actions 工作流变更——没有 CLI 行为可在 tmux 中驱动。行为回放测试即为真实场景验证：它们从工作流 YAML 中逐字提取 jq 过滤器和 bash 资格逻辑，对夹具回放。
关键行为覆盖：fork 候选并集 jq 准入 bot 无标签 fork（#20）和带标签人类 fork（#9），丢弃无 allow-edits 的 bot fork（#19）、in-repo bot PR（#18）和 skip fork（#7）。资格回放确认带 allow-edits 的 bot fork 无标签即合格，无 allow-edits 则丢弃。 — Qwen Code · qwen3.7-max Reviewed at |
Collaborator
|
Confidence: 5/5 — clean across every stage; the problem is real, the fix is minimal, and the safety envelope is preserved. This is a textbook small PR: two surgical edits that remove a redundant label requirement for the bot's own fork PRs, with matching test coverage. The independent proposal I sketched before reading the diff was identical to what landed — union the bot's forks into the scan pool, relax the fork-chain condition for bot authors. No simpler path exists; the change is already at the minimum. What I verified:
If I had to maintain this in six months, I'd thank the author — the comments explain the two-source union clearly, and the test fixtures document the admission/discard matrix. 中文说明置信度：5/5 — 每个阶段都干净；问题真实、修复最小、安全边界完整保留。 这是一个教科书级的小 PR：两处精准修改，移除对 bot 自己 fork PR 的冗余标签要求，配套测试覆盖。我在读 diff 之前草拟的独立方案与最终实现完全一致——将 bot fork 并入扫描池、放宽 bot 作者的 fork 链条件。不存在更简路径；变更已是最小。 验证内容：
— Qwen Code · qwen3.7-max Reviewed at |
qwen-code-ci-bot
approved these changes
Jul 19, 2026
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
Collaborator
There was a problem hiding this comment.
LGTM, looks ready to ship. ✅
Contributor
Code Coverage Summary
CLI Package - Full Text ReportCore Package - Full Text ReportFor detailed HTML reports, please see the 'coverage-reports-22.x-ubuntu-latest' artifact from the main CI run. |
yiliang114
approved these changes
Jul 19, 2026
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
Collaborator
There was a problem hiding this comment.
Reviewed — no blockers. Suggestions are inline.
— qwen3.8-max-preview via Qwen Code /review
Comment on lines
+672
to
+676
| expect(runRecheck(botFork).passed).toBe(true); | ||
| // Remove allow-edits and the same bot fork discards (cannot push). | ||
| expect(runRecheck({ ...botFork, maintainerCanModify: false }).passed).toBe( | ||
| false, | ||
| ); |
Collaborator
There was a problem hiding this comment.
[Suggestion] No negative test guards the complement of the new bot exemption — a NON-bot fork with allow-edits enabled but without the autofix/takeover label. botFork proves the bot passes without the label and forkPr proves a human passes with it, but nothing asserts that a human fork without the label is still rejected when allow-edits is on. — Concrete cost: if the author gate above (line 2053) is ever relaxed, a non-bot fork could be admitted with no failing test to catch it.
Suggested change
| expect(runRecheck(botFork).passed).toBe(true); | |
| // Remove allow-edits and the same bot fork discards (cannot push). | |
| expect(runRecheck({ ...botFork, maintainerCanModify: false }).passed).toBe( | |
| false, | |
| ); | |
| expect(runRecheck(botFork).passed).toBe(true); | |
| // Remove allow-edits and the same bot fork discards (cannot push). | |
| expect(runRecheck({ ...botFork, maintainerCanModify: false }).passed).toBe( | |
| false, | |
| ); | |
| // A NON-bot fork with allow-edits but no takeover label still discards — | |
| // the bot exemption is author-specific. | |
| expect( | |
| runRecheck({ ...botFork, author: { login: 'maint-fork' } }).passed, | |
| ).toBe(false); |
— qwen3.8-max-preview via Qwen Code /review
| # elif ends the ladder for eligible forks, so anything below it | ||
| # would be unreachable for exactly the PR class we fetch and push. | ||
| elif [[ "${LIVE_XREPO}" != "false" && "${LIVE_TAKEOVER}" != "true" ]]; then INELIGIBLE='fork head without takeover' | ||
| elif [[ "${LIVE_XREPO}" != "false" && "${LIVE_TAKEOVER}" != "true" && "${LIVE_AUTHOR}" != "${AUTOFIX_BOT}" ]]; then INELIGIBLE='fork head without takeover (a non-bot fork needs the label)' |
Collaborator
There was a problem hiding this comment.
[Suggestion] This fork-specific elif is now provably unreachable. The author gate above (line 2053, LIVE_AUTHOR != AUTOFIX_BOT && LIVE_TAKEOVER != "true") already rejects every PR that could satisfy this condition: any PR reaching here has author == bot OR takeover == true, which contradicts this branch's author != bot && takeover != true. The bot exemption added by this PR removed the last reachable case (previously bot forks without the label were caught here). — Concrete cost: the diagnostic 'fork head without takeover (a non-bot fork needs the label)' can never appear in a log (a non-bot fork is reported from line 2053 with no fork indication), and a future maintainer editing fork-label policy here would see no effect since the real gate is line 2053. If kept as defense-in-depth, say so in a comment so it isn't mistaken for an accidental dead branch.
Suggested change
| elif [[ "${LIVE_XREPO}" != "false" && "${LIVE_TAKEOVER}" != "true" && "${LIVE_AUTHOR}" != "${AUTOFIX_BOT}" ]]; then INELIGIBLE='fork head without takeover (a non-bot fork needs the label)' | |
| # Non-bot forks without takeover are already rejected by the author | |
| # gate above; this branch is unreachable defense-in-depth. |
— qwen3.8-max-preview via Qwen Code /review
Collaborator
|
Released in v0.20.1. |
Collaborator
|
Agent run timed out after 1800000ms ❌ failed |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this PR does
Auto-manages the autofix bot's own fork PRs without requiring a label.
The bot's codex flow opens PRs from its own fork (
qwen-code-dev-bot/qwen-code). Those are the bot's own generated work — same author, same code provenance, and the bot holdswrite+— so they are trust-equal to an in-repo bot PR, which is already auto-managed with no label. Requiring a manualautofix/takeoverlabel on them was redundant: the takeover label exists to authorize external (human) fork authors, not the bot itself. That's exactly why #7208 and #7220 (bot codex-flow fork PRs with actionable review suggestions) sat unprocessed until a label was applied by hand.Now a fork authored by the autofix bot with "Allow edits from maintainers" is admitted and managed without a label:
bot-prs.json(the bot's own forks — the list is--author AUTOFIX_BOT, so no label needed) and the takeover-labeled list (non-bot forks, explicit opt-in). Both still require allow-edits and pass the per-candidate livewrite+gate.write+author, and a matching live head repo.Unchanged:
autofix/skipstill opts any such PR out; non-bot forks still need the explicitautofix/takeoverlabel.Why it's safe
A bot-authored fork adds no trust surface beyond an in-repo bot PR:
authorcannot be spoofed (it is the account that opened the PR), the fork is the bot's own (only the bot pushes to it), the executed-code provenance is identical to in-repo bot PRs, and the address run's constraints (coreTools allowlist, docker sandbox, severed hooks, trusted-base runner staging) are unchanged. Thewrite+gate and allow-edits requirement remain.Reviewer Test Plan
How to verify
npx vitest run scripts/tests/qwen-autofix-workflow.test.js— 62/62. Behavioral coverage:npx vitest run scripts/tests/qwen-fleet-shepherd-workflow.test.js— 12/12.run:block passesbash -n.Tested on
Risk & Scope
autofix/skipis the opt-out; allow-edits + livewrite+remain hard gates.Linked Issues
Follow-up to #7213 (fork takeover). Motivated by #7208 / #7220 sitting unprocessed for want of a label.
中文说明
本 PR 做了什么
让 autofix bot 自己 fork 的 PR 无需标签即自动托管。
bot 的 codex 流程会从自己的 fork(
qwen-code-dev-bot/qwen-code)开 PR。那是 bot 自己生成的工作 —— 同一作者、同样的代码来源、bot 持有write+—— 与 in-repo bot PR 同等信任,而后者本就无需标签自动托管。给它们打autofix/takeover标签是多余的:该标签是给外部(人类) fork 作者的显式授权,不是给 bot 自己的。这正是 #7208、#7220(带可执行 review suggestion 的 bot codex fork PR)一直没被处理、直到手动打标签的原因。现在:由 autofix bot 作者、且勾选 "Allow edits from maintainers" 的 fork,无需标签即被纳入并托管:
bot-prs.json(bot 自己的 fork —— 列表是--author AUTOFIX_BOT,无需标签)与 takeover 标签列表(非 bot fork,显式 opt-in)并集取得;两者仍需 allow-edits 并逐候选过实时write+门。write+作者 + 实时 head 仓库匹配。不变:
autofix/skip仍可让任意此类 PR 退出;非 bot fork 仍需显式autofix/takeover标签。为什么安全
bot 作者的 fork 不增加超出 in-repo bot PR 的信任面:
author无法伪造(就是开 PR 的账号)、fork 是 bot 自己的(只有 bot 推送)、被执行代码来源与 in-repo bot PR 一致、address 运行的约束(coreTools 白名单、docker 沙箱、断 hook、可信基座 runner staging)均不变。write+门与 allow-edits 要求保留。评审验证
npx vitest run scripts/tests/qwen-autofix-workflow.test.js—— 62/62:fork 候选并集 jq(逐字提取、双夹具回放)准入 bot 无标签 fork([Bug] API请求超时：无法完成html重构（啥活没产，就干了我input 43万个tokens） #20)与带标签人类 fork(API Error: Streaming setup timeout after 45s #9),丢弃无 allow-edits 的 bot fork、in-repo bot PR、skip fork;资格回放使带 allow-edits 的 bot fork 无标签即合格、无 allow-edits 则丢弃。npx vitest run scripts/tests/qwen-fleet-shepherd-workflow.test.js—— 12/12。run:块过bash -n。风险与范围
autofix/skip为退出开关,allow-edits + 实时write+仍是硬门。关联 Issue
#7213(fork 托管)的后续;由 #7208 / #7220 因缺标签未被处理驱动。