How shivansh19200/github-repository-reviewer works
This is a GitHub repository review tool. Its visible review workflow runs an architecture review and then a code-quality review over repository files held in shared graph state; each agent asks an LLM to review file chunks and writes a report back to that state. Separate modules parse GitHub URLs and fetch and preprocess repository contents, but the sampled code does not connect those modules to the review graph or show an application entry point. The README is empty, so broader capabilities cannot be confirmed.
Main entry points: architecture.py Architecture agent · code_quality_agent.py Code quality agent · graph.py Review graph · preprocessor.py Preprocessor
Components 9
Review workflow
- Review graph
graph.py
Repository ingestion
- URL parser
url_parser.py - Preprocessor
preprocessor.py - Repository indexer
indexer.py
AI analysis
- Architecture agent
agents/architecture.py - Code quality agent
agents/code_quality_agent.py
Other components
- Reviewer
- GitHub API
- OpenAI model
Connections 9
- Review graph → Architecture agent: dispatches
- Review graph → Code quality agent: dispatches
- Architecture agent → Code quality agent: then dispatches
- Architecture agent → OpenAI model: requests reviews
- Code quality agent → OpenAI model: requests reviews
- URL parser → GitHub API: looks up branch
- Preprocessor → Repository indexer: gets file paths
- Preprocessor → GitHub API: fetches contents
- Repository indexer → GitHub API: fetches tree