Skip to content

chore: add check-models skill + CI gate (tooling only)#67

Merged
jimbobbennett merged 2 commits into
mainfrom
chore/check-models-skill
Jun 22, 2026
Merged

chore: add check-models skill + CI gate (tooling only)#67
jimbobbennett merged 2 commits into
mainfrom
chore/check-models-skill

Conversation

@jimbobbennett

@jimbobbennett jimbobbennett commented Jun 18, 2026

Copy link
Copy Markdown
Contributor

Adds the check-models skill and a CI gate that flags outdated OpenAI / Anthropic / Google (Gemini) model references.

Scope: tooling only — 4 files. The CI gate (check-models.yml) scans only the lines a PR changes (--diff <base>), so it fails the check when a PR introduces an outdated model ID on a changed line, without blocking on pre-existing references elsewhere.

The bulk model-reference migrations (done and live-verified separately) are parked on the chore/check-models branch and will land as small, themed follow-up PRs (gpt-5, Claude, gpt-4.1, phoenix.evals API, Gemini) for easier review.

To see this gate in action, here is a throwaway demo PR (now closed) that edited one line containing an outdated model:

The check-models check failed with 1 introduced (the touched line → gpt-4o-minigpt-5.4-mini) and listed 6 pre-existing outdated models elsewhere in the same file as non-blocking warnings. The bot comment on #68 shows the two tiers.

jimbobbennett and others added 2 commits June 18, 2026 15:46
Skill + scanner that find outdated OpenAI / Anthropic / Google (Gemini)
model references and migrate them to current size-tier equivalents. The
CI gate scans only the lines a PR changes (paths: python typescript), so
it flags newly introduced outdated models and fails the check, without
blocking on pre-existing references. Notebook/code migrations land
separately as themed PRs.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…sting

CI gate now scans the whole of each touched file and tags findings by whether
the PR changed the line — introduced fails, pre-existing (unchanged line of a
touched file) is a non-blocking warning. `python typescript` scan paths.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@jimbobbennett jimbobbennett marked this pull request as ready for review June 18, 2026 23:09
@jimbobbennett jimbobbennett merged commit e633917 into main Jun 22, 2026
1 check passed
@jimbobbennett jimbobbennett deleted the chore/check-models-skill branch June 22, 2026 23:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants