Run an AI code review on any public pull request in under 5 minutes. No account needed — just bring your own API key.
ANTHROPIC_API_KEY) or Claude Code OAuth token.gh/glab CLI logged in (token setup, CLI setup)gh or glab CLI is installed and logged in, Reviewate uses it automatically. Docker and CI always require a token.pip install reviewate
# From a URL (auto-detects platform, repo, PR):
reviewate https://github.com/facebook/react/pull/28000
# Or with owner/repo + PR number:
reviewate facebook/react -p 28000
# First run: prompts for model choices, saves to ~/.reviewate/config.toml
Also works with uv tool install reviewate.
# Clone and install
git clone https://github.com/numberly/reviewate.git
cd reviewate/code_reviewer
uv sync
# Run — first-run setup will prompt for model choices
uv run python main.py https://github.com/facebook/react/pull/28000
# From PR/MR URL (auto-detects platform, repo, PR number):
reviewate https://github.com/org/repo/pull/48
reviewate https://gitlab.com/group/repo/-/merge_requests/1
# Classic format:
reviewate owner/repo -p 123 # review (default)
reviewate summary owner/repo -p 123 # summary only
reviewate full owner/repo -p 123 # review + summary
reviewate config # re-run setup wizard
| Flag | Description |
|---|---|
-p / --pr | Pull request / merge request number (auto-detected from URL) |
--platform | github or gitlab (auto-detected from URL) (default is github) |
--dry-run | Run without posting comments to the platform |
--debug | Enable debug mode with tracebacks and reasoning output |
--json | Output results as JSON |
On first run, Reviewate prompts you to choose a model and saves your config to ~/.reviewate/config.toml:
[models]
review = "sonnet"
utility = "haiku"
API keys are read from environment variables only — they are never stored in the config file.
You can also configure via environment variables — they override config file values:
export REVIEWATE_REVIEW_MODEL=sonnet
export REVIEWATE_UTILITY_MODEL=haiku
Override priority: env vars > config file > defaults
See Model Configuration for details on the two-tier model system.