🤖 feat: refresh LiteLLM models and prune stale overrides#2559
🤖 feat: refresh LiteLLM models and prune stale overrides#2559ammar-agent wants to merge 6 commits intomainfrom
Conversation
|
@codex review Please review the models refresh + models-extra pruning + update-models skill improvements. |
ammar-agent
force-pushed
the
feat/update-models-cycle
branch
from
a8a75e1 to
b84b1c1
Compare
|
@codex review Rebased onto latest main, resolved merge conflicts, and re-ran static-check + targeted model tests. |
|
Codex Review: Didn't find any major issues. You're on a roll. ℹ️ About Codex in GitHubCodex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback". |
|
@codex review Addressed the failing integration check by making compaction UI tests deterministic with an explicit lower-context model for compaction flows. Re-ran static-check and targeted tests. |
![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: fab0f4d20f
ℹ️ About Codex in GitHub
Codex has been enabled to automatically 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 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
@codex review Addressed the draft-preservation test concern by keeping sendMessage in-flight while typing the draft, then awaiting completion. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8b5999184e
ℹ️ About Codex in GitHub
Codex has been enabled to automatically 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 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
@codex review Adjusted the compaction draft-preservation test to keep the UI send path ( |
|
Codex Review: Didn't find any major issues. 🚀 ℹ️ About Codex in GitHubCodex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback". |
ammar-agent
force-pushed
the
feat/update-models-cycle
branch
from
d4a0506 to
d5d8af9
Compare
|
@codex review |
|
Codex Review: Didn't find any major issues. Breezy! ℹ️ About Codex in GitHubCodex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback". |
1 participant
Summary
This PR adds a reusable
update-modelsskill, runs a full model-refresh cycle, and prunes stalemodels-extraoverrides that upstream LiteLLM now covers.Background
models-extra.tsis checked beforemodels.json, so stale local overrides silently shadow corrected upstream metadata. We needed a repeatable workflow that makes pruning safe and routine.Implementation
.mux/skills/update-models/SKILL.mdwith an end-to-end workflow for:models.jsonmodels-extraagainst upstreamsrc/common/utils/tokens/models.jsonfrom upstream LiteLLM.src/common/utils/tokens/models-extra.tsdown to only models still missing upstream (gpt-5.3-codex,gpt-5.3-codex-spark).tests/ui/compaction/compaction.test.tsby pinning compaction-flow test sends to Sonnet so auto-compaction still has a deterministic higher-context fallback after Opus metadata moved to 1M context upstream.Validation
make static-checkbun test src/common/constants/knownModels.test.ts src/common/utils/tokens/modelStats.test.ts src/common/utils/ai/modelCapabilities.test.tsTEST_INTEGRATION=1 bun x jest tests/ui/compaction/compaction.test.ts --runInBandRisks
Generated with
mux• Model:openai:gpt-5.3-codex• Thinking:xhigh• Cost:$2.12