[PR #21] [MERGED] fix(cursor-review): bump dead kimi-k2.5 model id to kimi-k2.7-code #31

New issue

Closed

opened 2026-10-26 22:35:35 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-26 22:35:35 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/Comfy-Org/github-workflows/pull/21
Author: @mattmillerai
Created: 7/7/2026
Status: ✅ Merged
Merged: 7/7/2026
Merged by: @mattmillerai

Base: main ← Head: matt/bump-kimi-k2.7-code

📝 Commits (1)

b0af3d5 fix(cursor-review): bump dead kimi-k2.5 model id to kimi-k2.7-code

📊 Changes

2 files changed (+2 additions, -2 deletions)

View changed files

📝 .github/cursor-review/README.md (+1 -1)
📝 .github/workflows/cursor-review.yml (+1 -1)

📄 Description

ELI-5

Our PR-review bot asks four AI models to look at each change. One of them, Kimi, got renamed on Cursor's side, so it's been silently coming back blank — we've quietly been reviewing with three models instead of four. This swaps in the current Kimi model so all four vote again.

What changed

Panel matrix model kimi-k2.5 → kimi-k2.7-code in .github/workflows/cursor-review.yml.
Model table in .github/cursor-review/README.md updated to match.

Why

cursor-agent --model kimi-k2.5 now exits 1 with empty stdout and stderr Cannot use this model: kimi-k2.5. The run step's || true swallows the exit code, so extract-findings.py sees empty stdout and scores the cell empty — indistinguishable from "the model ran and found nothing." Both Kimi cells (adversarial + edge-case) drop out on every run and show up as Reviewers that did not contribute.

kimi-k2.7-code is the only Kimi model left in Cursor's catalog (verified live — it returns output normally). The other three pins — gpt-5.3-codex-xhigh, claude-opus-4-8-thinking-xhigh, gemini-3.1-pro — are all still current and unchanged.

Follow-up

A separate change will make an unknown/delisted model id fail loud (a catalog preflight, or scoring the cell error instead of empty) so the next time Cursor rotates its catalog it surfaces immediately instead of quietly degrading the panel.

🤖 Generated with Claude Code

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/Comfy-Org/github-workflows/pull/21 **Author:** [@mattmillerai](https://github.com/mattmillerai) **Created:** 7/7/2026 **Status:** ✅ Merged **Merged:** 7/7/2026 **Merged by:** [@mattmillerai](https://github.com/mattmillerai) **Base:** `main` ← **Head:** `matt/bump-kimi-k2.7-code` --- ### 📝 Commits (1) - [`b0af3d5`](https://github.com/Comfy-Org/github-workflows/commit/b0af3d56b324d255569bfdc59cf7e07d9a7e9a40) fix(cursor-review): bump dead kimi-k2.5 model id to kimi-k2.7-code ### 📊 Changes **2 files changed** (+2 additions, -2 deletions) <details> <summary>View changed files</summary> 📝 `.github/cursor-review/README.md` (+1 -1) 📝 `.github/workflows/cursor-review.yml` (+1 -1) </details> ### 📄 Description ## ELI-5 Our PR-review bot asks four AI models to look at each change. One of them, Kimi, got renamed on Cursor's side, so it's been silently coming back blank — we've quietly been reviewing with three models instead of four. This swaps in the current Kimi model so all four vote again. ## What changed 1. Panel matrix model `kimi-k2.5` → `kimi-k2.7-code` in `.github/workflows/cursor-review.yml`. 2. Model table in `.github/cursor-review/README.md` updated to match. ## Why `cursor-agent --model kimi-k2.5` now exits 1 with empty stdout and stderr `Cannot use this model: kimi-k2.5`. The run step's `|| true` swallows the exit code, so `extract-findings.py` sees empty stdout and scores the cell `empty` — indistinguishable from "the model ran and found nothing." Both Kimi cells (adversarial + edge-case) drop out on every run and show up as `Reviewers that did not contribute`. `kimi-k2.7-code` is the only Kimi model left in Cursor's catalog (verified live — it returns output normally). The other three pins — `gpt-5.3-codex-xhigh`, `claude-opus-4-8-thinking-xhigh`, `gemini-3.1-pro` — are all still current and unchanged. ## Follow-up A separate change will make an unknown/delisted model id fail loud (a catalog preflight, or scoring the cell `error` instead of `empty`) so the next time Cursor rotates its catalog it surfaces immediately instead of quietly degrading the panel. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-26 22:35:35 +00:00