Conversation
- Rename Auto model label to a descriptive tagline so /model shows what it does instead of internal branding - Add `auto (<model id>)` row to prompt summaries once the router resolves a target, mirroring the status bar segment - Announce the router pick via a status-line notice when the decision happens instead of silently recording it Co-Authored-By: Kimchi <noreply@kimchi.dev>
Pre-push hook was failing on three files; auto-fixed with biome. Co-Authored-By: Kimchi <noreply@kimchi.dev>
Kimchi Code Review
Summary📊 Review Score: 84/100 (overall code quality — 0 lowest, 100 highest) 🧪 Tests: yes — Good coverage overall: new unit tests in 📝 Found 6 issue(s). See inline comments for details. What to expectKimchi will analyze the changes in this pull request and post:
The review typically completes within a few minutes. This comment will be updated once the review is ready. Interact with Kimchi
ConfigurationReviews are configured by your organization admin. Powered by Kimchi — AI-powered code review by CAST AI |
![kimchi-review[bot]](https://avatars.githubusercontent.com/in/3223788?s=60&v=4){kind=small}
There was a problem hiding this comment.
📊 Review Score: 84/100 (overall code quality — 0 lowest, 100 highest)
⏱️ Estimated effort to review: 2/5 (1 = trivial, 5 = very complex)
🧪 Tests: yes — Good coverage overall: new unit tests in prompt-summary.test.ts cover the resolved, unresolved, and concrete-model cases for the new model row; router/index.test.ts was updated to assert the new notify call; models.test.ts and the e2e /model autocomplete test were updated for the rename. Minor gap: the multi-row (subagent) renderer branch that also appends the model row has no test.
📝 Found 6 issue(s). See inline comments for details.
Address PR #1221 review feedback: - Extract formatAutoModelLabel() into router/state.ts so the status bar's model segment and the prompt summary's model row share one source for the `auto (<model id>)` format instead of open-coding it - Hoist the duplicated model-row block in the prompt-summary renderer so both the single-row and multi-row branches share one implementation - Add a multi-row (subagent breakdown) renderer test asserting the model row is placed after the per-model rows Co-Authored-By: Kimchi <noreply@kimchi.dev>
The Auto-model pick notice was shown with a transient ctx.ui.notify, so it vanished on session resume even though the decision itself persisted in the kimchi_auto_resolution custom entry. Render that entry via pi.registerEntryRenderer instead: the same status line now appears when the decision is made and on every replay, reusing data already on disk, and custom entries stay out of LLM context (unlike sendMessage). Co-Authored-By: Kimchi <noreply@kimchi.dev>
1 participant
What does this PR do?
what it does instead of internal branding
auto (<model id>)row to prompt summaries once the routerresolves a target, mirroring the status bar segment
decision happens instead of silently recording it