Show HN: CodingFox – Open-Source AI Code Review Tool That Works Like Magic (github.com)

CodingFox is an open‑source AI code review assistant that plugs into GitHub Actions (codingfox/ai-pr-reviewer@latest) to perform instant, contextual PR analysis using OpenAI models (GPT‑3.5 Turbo for summaries and GPT‑4 for in‑depth reviews). It offers line‑by‑line suggestions, automated PR summaries/release notes, incremental per‑commit reviews, pattern/anti‑pattern detection, security flagging, test generation, chat-style queries on code, and customizable prompts and path filters. Setup requires adding an OPENAI_API_KEY secret, dropping a provided .github/workflows/codingfox-review.yml into your repo, and meeting minimal Node.js tooling — the repo is MIT‑licensed and includes options to tune verbosity, comment limits, concurrency, and model selection. For the AI/ML and dev‑tools community, CodingFox showcases a practical, configurable LLM integration into developer workflows: it can dramatically speed reviews (projected 60–65% time reduction) and surface more pre‑production bugs (vendor claims ~40% more), while balancing cost (est. ~$0.002 per PR with GPT‑3.5 vs ~$0.10–0.50 with GPT‑4). Key implications include real productivity gains, opportunities for continuous learning and team‑specific prompt engineering, and tradeoffs around data privacy and compliance because code is sent to OpenAI’s API (on‑premise options are mentioned). It’s a useful reference implementation for applying large language models to code understanding, but teams should weigh cost, rate limits, and sensitive‑data policies before wide deployment.