juniqe-com/Pi Code Review
AI-powered code review using Pi. Posts inline comments on PR files.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| model | Single model in provider/model format. Ignored when 'models' is set. | no | "" |
| models | Comma-separated list of models to randomly choose from (e.g. 'openai/gpt-5.4, anthropic/claude-sonnet-4-20250514'). One is picked at random per run. | no | "" |
| thinking | Optional thinking level (e.g. off, minimal, low, medium, high, xhigh) | no | "" |
| github_token | GitHub token for posting review comments. Defaults to the built-in GITHUB_TOKEN. | no | ${{ github.token }} |
| pi_version | Pi version to install. Defaults to a pinned, known-good version. | no | 0.80.10 |
| review_prompt | Custom review prompt to append to the default instructions. Use this to add project-specific review criteria. | no | "" |
| max_diff_size | Maximum diff size in bytes before truncation (default: 100000) | no | 100000 |
| review_timeout | Seconds to wait for Pi before stopping the review. Pi writes findings incrementally, so on timeout the action still posts whatever it gathered as a partial review, then fails the action. Default 900 (15 min) is ~2x observed median, well above p90 — catches hangs without killing legitimately slow runs. | no | 900 |
| extra_env | Extra environment variables to pass to Pi as KEY=VALUE lines | no | "" |
Outputs
no outputs