docs(devlog): close the Devin unit and record why AssignModel is not implemented - #4464
Conversation
…implemented Six changes landed on dev from this unit. The seventh, an AssignModel handshake for router uids, is closed as a no-op because an independent audit showed it would be dead code on the request path. The case for it was that the native CLI and CLIProxyAPIPlus both call that RPC and we do not. The first half is true. The second half does not follow: Plus gates the call behind a -router suffix or model-router substring, and neither pattern appears in our catalog parse, our static model list, or the binary survey. The adaptive uid that prompted the question comes from GetCliModelConfigs, and we parse GetCascadeModelConfigs. Even granting a router uid, the hook has nowhere to run. resolveWireModelUid rewrites a bare adaptive to adaptive-medium, and the catalog preflight then ends the turn with not_listed, both before the point where AssignModel would be called. Moving the hook earlier would mean reversing the suffix handling just unified and the preflight contract from #14, to serve demand nobody has observed. The verified envelope is kept in 060 so the phase can reopen cheaply if a router uid is ever seen, along with the two conditions that would reopen it. c-8 stays unmet rather than reworded. The criterion was written on a premise that turned out to be false, and weakening it to close the unit would make the record useless. The closeout also records the four places where audit changed the plan, which is the part worth reading: an abort reason that would have been swallowed by AbortError, a global cache clear that would have cut other accounts mid-turn, an epoch that was dead complexity, and a tier token that would have become an accepted reasoning effort. Planning only. Local product suite, typecheck, build and install: NOT RUN.
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
github-actions
Bot
added
the
documentation
github-actions
Bot
added
the
intake: hygiene-blocked
|
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: 📒 Files selected for processing (2)
Included review availability: Your plan provides up to 10 included reviews per hour; 5 remain after this review. 📝 WalkthroughWalkthroughThe PR adds final records for the 260913 workunit and wp6. It documents merged PRs, CI evidence, audit-driven plan changes, deferred items, and the decision to leave AssignModel unmet and closed as a NOOP. ChangesWorkunit closure
Priority: ⬇️ Low Estimated code review effort: 1 (Trivial) | ~5 minutes Change: Other Merge Risk: ⚪ Minimal · up to The documentation accurately records the audited no-op decision and its reopening conditions. No actionable merge risk remains. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches 💡 1🛠️ Fix failing CI checks 💡
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
✅ READY
Hygiene✅ Deterministic PR hygiene checks passed. |
Pull request was converted to draft
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 16bf77ad3b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| ## 단위 종료 기록 (2026-09-13) | ||
|
|
||
| 6개 PR이 `dev`에 들어갔고, 1개 단계는 NOOP으로 닫혔다. |
There was a problem hiding this comment.
_fin
This section records a terminal outcome and explicitly closes the unit, but all seven documents remain under devlog/_plan/, so the repository's planning queue continues to classify completed work as open. Move the entire 260913_devin_landing_and_caching directory to devlog/_fin/ as part of this closeout.
AGENTS.md reference: AGENTS.md:L102-L105
Useful? React with 👍 / 👎.
리뷰 · 우선순위 42 / 80설명 이 PR은 경로 github-actions hygiene - 지금 게이트가 경로 devlog/_plan/260913_devin_landing_and_caching/000_plan.md - 단위 종료 표·감사 표·미해결( 경로 …/060_wp6_assign_model_router.md - NOOP 결론·블로커 2개·재개 조건이 요청 경로 현실과 맞습니다. 제품 코드 추가 없이 기록을 남기는 선택이 옳아 보입니다. 메인테이너의 판단이 필요한 지점
너의 추천 이 댓글은 grok-bot이 작성했습니다 |
github-actions
Bot
removed
the
intake: hygiene-blocked
|
Maintainer integration into
|
1 participant
Summary
Closes the
260913_devin_landing_and_cachingunit. Six changes landed ondevfrom it; the seventh is closed as a no-op and this records why.The case for an AssignModel handshake was that the native CLI and CLIProxyAPIPlus both call that RPC and we do not. The first half is true. The second half does not follow — Plus gates the call behind a
-routersuffix ormodel-routersubstring, and neither pattern appears in our catalog parse, our static model list, or the binary survey. Theadaptiveuid that prompted the question comes fromGetCliModelConfigs; we parseGetCascadeModelConfigs.Even granting a router uid, the hook has nowhere to run.
resolveWireModelUidrewrites a bareadaptivetoadaptive-medium, and the catalog preflight then ends the turn withnot_listed— both before the point where AssignModel would be called. Moving the hook earlier would mean reversing the suffix handling just unified and the preflight contract from #14, to serve demand nobody has observed.The verified request/response field layout is kept in
060so the phase reopens cheaply if a-routeruid is ever seen in aGetCascadeModelConfigsresponse, or a turn is observed dying asnot_listedon one.What landed
0a89b416aeee8fd82f720ea9730dc33113a9261bab915cff737ce4c-8stays unmet rather than reworded. It was written on a premise that turned out to be false, and weakening it to close the unit would make the record useless.Verification
failwith the two blockers above. The re-audit confirmed they are restated without softening, that no cheap correct middle exists (a named error would require either treatingadaptiveas a router on no evidence, or a dead branch on an unobserved pattern), that the closeout merge SHAs are real ancestors oforigin/dev, and that the unmet criterion was not reworded to pass.invalid_argumentburns a credential cooldown was carried over from the design doc and overstated, since key-failover cooldown only fires on 401/429. Corrected in place.Checklist
Maintainer integration under
MAINTAINERS.md:devonly, with exact-head CI evidence recorded before merge.This PR records the landing of work by two other authors. Their credit is on the squashed commits of #4445 and #4448; repeated here so the attribution gate sees it on the summarising PR as well.
Co-authored-by: Smartnewb 159137930+Smartnewb@users.noreply.github.com
Co-authored-by: luvs01 27862058+luvs01@users.noreply.github.com