Conversation
Remove the CLI fallback that credited missing legacy modelUsage to the currently selected UI model. Unknown attribution is skipped rather than fabricated. Update the usage delivery Spec and Agent Notes for Codex exact per-response attribution, the Codex adapter change, and the durable resume boundary. Model: deepseek-flash
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. |
![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: bb0052d8d1
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".
| - Never synthesize `modelUsage` from `currentModel`. A legacy adapter that omits model | ||
| attribution stays skipped/unattributed rather than being credited to the UI selection. |
There was a problem hiding this comment.
This addition grows apps/cli/src/agent/AGENTS.md from 8,167 to 8,343 bytes, exceeding the repository's 8 KiB hard limit and causing the required documentation check to reject the change. Move explanatory content to the owning note or shorten the scoped rules before landing.
AGENTS.md reference: AGENTS.md:L26-L27
Useful? React with 👍 / 👎.
Record unresolved adapter P1s and measured ablation results. Preserve CLI projection and document missing-dependency validation limits. Model: gpt-6
1 participant
Risk/Confidence: medium/medium
Summary
modelUsageto the currently selected UI model.Dependency and review status
The adapter implementation is LodyAI/acp-extension-codex#45. Independent review plus ablation found unresolved adapter P1s, so #45 is explicitly not ready to merge and the submodule pointer bump remains deferred. This PR intentionally does not bump the nested submodule pointer.
Verification
esbuildbundle forapps/cli/src/agent/agent-client.tspasses.git diff --checkpasses.pnpm check/Vitest was not run in this worktree because root dependencies are not installed.Model: deepseek-flash