AI Code Review Assistant
A GitHub-integrated code review tool that runs local language models through Ollama to analyze pull requests. Provides context-aware suggestions, security vulnerability detection, and style consistency checks while keeping code entirely private.
Key Highlights
- Local LLM inference via Ollama — no data leaves your infrastructure
- GitHub Actions integration for automated PR review triggers
- Multi-model support with model selection based on review type
- Security-focused analysis with CVE pattern matching
- FastAPI backend for custom review pipeline configuration
Architecture
A GitHub Action workflow triggers on PR events, sending diff data to a FastAPI service that coordinates with Ollama for LLM inference. The system uses a chain-of-thought prompting strategy: classify the change type → analyze for issues → formulate suggestions → rank by severity.
Other Projects
Light Novel Tracker Notifier
Multi-source scraping engine that monitors light novel releases across platforms and delivers real-time notifications via a scalable polling architecture.
Flask SaaS Boilerplate
Production-ready Flask starter kit with authentication, billing, team management, and deployment configuration for Railway.