LLM-powered code review tool that investigates your code for bugs, security issues, and code smells. Supports multiple LLM providers and runs as a CLI or GitHub Action.
- Multi-provider — OpenAI, Anthropic, Google Gemini, Groq, and local models via Ollama
- Smart Analysis — detects bugs, security vulnerabilities, code smells, and logic errors
- Git-aware — review staged changes, diffs between branches, or entire directories
- Configurable — severity levels, ignore patterns, custom rules via
.benno.yaml - GitHub Action — automated PR review with inline comments
- Fast — parallel file analysis with configurable concurrency
pip install detective-benno# review a file or directory
benno investigate src/main.py
benno investigate src/
# review staged changes
benno staged
# review a branch diff
git diff main..feature | benno diffbenno investigate --provider ollama --model codellama src/
benno investigate --provider anthropic src/
benno investigate --provider gemini src/Run benno init to create a .benno.yaml:
provider: openai
model: gpt-4o
severity: medium # minimum severity to report
ignore:
- "*.test.go"
- "vendor/"
concurrency: 4- uses: bimakw/detective-benno@v1
with:
github_token: ${{ secrets.GITHUB_TOKEN }}
openai_api_key: ${{ secrets.OPENAI_API_KEY }}
severity: mediumpytestApache 2.0 — see LICENSE.