[PR #378] [MERGED] cursor-review: lower OpenAI lane to gpt-5.6-sol-xhigh #380

New issue

Closed

opened 2026-10-26 22:39:12 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-26 22:39:12 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/Comfy-Org/github-workflows/pull/378
Author: @claude[bot]
Created: 10/8/2026
Status: ✅ Merged
Merged: 10/8/2026
Merged by: @mattmillerai

Base: main ← Head: lower-sol-lane-to-xhigh

📝 Commits (1)

cfdf5ed cursor-review: lower the OpenAI lane from gpt-5.6-sol-max to gpt-5.6-sol-xhigh

📊 Changes

4 files changed (+13 additions, -5 deletions)

View changed files

📝 .github/cursor-review/README.md (+5 -1)
📝 .github/cursor-review/prompt-judge.md (+1 -1)
📝 .github/workflows/cursor-review.yml (+5 -1)
📝 docs/callers/cursor-review.md (+2 -2)

📄 Description

Requested by Matt Miller · Slack thread

Before: The OpenAI lane of the cursor-review panel ran gpt-5.6-sol-max, the highest effort tier. In recent review runs that lane frequently ran until the 15-minute cell cap and failed.

After: The OpenAI lane runs gpt-5.6-sol-xhigh, one effort tier lower, to reduce timeouts and run-to-cap failures on the sol lane. Everything else in the panel and judge is unchanged.

What it does: Swaps the built-in OpenAI panel pin from gpt-5.6-sol-max to gpt-5.6-sol-xhigh.

How: Changed the panel heredoc in .github/workflows/cursor-review.yml (plus a short comment), the panel table and a note in .github/cursor-review/README.md, the example in prompt-judge.md, and the CURSOR_PANEL_MODELS examples in docs/callers/cursor-review.md (they must differ from the built-in list by exactly one axis, so they now track the new pin). catalog-drift.py reads the pin out of the workflow, so it needs no change. Historical test fixtures (the captured 2026-08-10 catalog and the panels asserted against it) are intentionally left as captured.

Catalog check: gpt-5.6-sol-xhigh ("GPT-5.6 Sol 1M Extra High") is listed in the repo's captured catalog fixture .github/cursor-review/tests/fixtures/catalog-2026-08-10.txt. That is a snapshot, not a live fetch; preflight validates the id against the live catalog on every run and fails loudly if it is missing, and the weekly catalog-drift check would flag a delisting.

Verification: cursor-review test suite (1147 tests), check_workflow_pins.py, check_agents_md.py, and the org-repo-literals lint all pass locally. shellcheck was not available locally.

Rollout: This changes a run-time default in a reusable workflow, so callers pick it up when they bump their pin. No Skip-caller-bump trailer, deliberately.

🤖 Generated with Claude Code

https://claude.ai/code/session_01GR61xoEiciwzwPvV92KRRE

Generated by 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/378 **Author:** [@claude[bot]](https://github.com/apps/claude) **Created:** 10/8/2026 **Status:** ✅ Merged **Merged:** 10/8/2026 **Merged by:** [@mattmillerai](https://github.com/mattmillerai) **Base:** `main` ← **Head:** `lower-sol-lane-to-xhigh` --- ### 📝 Commits (1) - [`cfdf5ed`](https://github.com/Comfy-Org/github-workflows/commit/cfdf5ed054440c0900f8d26f2dbb334252c2b9af) cursor-review: lower the OpenAI lane from gpt-5.6-sol-max to gpt-5.6-sol-xhigh ### 📊 Changes **4 files changed** (+13 additions, -5 deletions) <details> <summary>View changed files</summary> 📝 `.github/cursor-review/README.md` (+5 -1) 📝 `.github/cursor-review/prompt-judge.md` (+1 -1) 📝 `.github/workflows/cursor-review.yml` (+5 -1) 📝 `docs/callers/cursor-review.md` (+2 -2) </details> ### 📄 Description  _Requested by **Matt Miller** · [Slack thread](https://comfy-organization.slack.com/archives/C0ALXPF3N73/p1791437995095669)_ **Before:** The OpenAI lane of the cursor-review panel ran `gpt-5.6-sol-max`, the highest effort tier. In recent review runs that lane frequently ran until the 15-minute cell cap and failed. **After:** The OpenAI lane runs `gpt-5.6-sol-xhigh`, one effort tier lower, to reduce timeouts and run-to-cap failures on the sol lane. Everything else in the panel and judge is unchanged. **What it does:** Swaps the built-in OpenAI panel pin from `gpt-5.6-sol-max` to `gpt-5.6-sol-xhigh`. **How:** Changed the panel heredoc in `.github/workflows/cursor-review.yml` (plus a short comment), the panel table and a note in `.github/cursor-review/README.md`, the example in `prompt-judge.md`, and the `CURSOR_PANEL_MODELS` examples in `docs/callers/cursor-review.md` (they must differ from the built-in list by exactly one axis, so they now track the new pin). `catalog-drift.py` reads the pin out of the workflow, so it needs no change. Historical test fixtures (the captured 2026-08-10 catalog and the panels asserted against it) are intentionally left as captured. **Catalog check:** `gpt-5.6-sol-xhigh` ("GPT-5.6 Sol 1M Extra High") is listed in the repo's captured catalog fixture `.github/cursor-review/tests/fixtures/catalog-2026-08-10.txt`. That is a snapshot, not a live fetch; preflight validates the id against the live catalog on every run and fails loudly if it is missing, and the weekly catalog-drift check would flag a delisting. **Verification:** cursor-review test suite (1147 tests), `check_workflow_pins.py`, `check_agents_md.py`, and the org-repo-literals lint all pass locally. shellcheck was not available locally. **Rollout:** This changes a run-time default in a reusable workflow, so callers pick it up when they bump their pin. No `Skip-caller-bump` trailer, deliberately. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01GR61xoEiciwzwPvV92KRRE --- _Generated by [Claude Code](https://claude.ai/code/session_01GR61xoEiciwzwPvV92KRRE)_ --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-26 22:39:12 +00:00