Hey HN, I built this because I was tired of spending 45 minutes grepping through CloudWatch at 3 AM trying to figure out why our backend was throwing 500s. Autopsy pulls your recent error logs and GitHub deploys, sends them to an LLM with a diagnostic prompt, and gives you a structured root cause analysis right in your terminal. Runs locally — logs go from AWS directly to the AI provider, nothing touches my servers. Built in Python, works with CloudWatch + GitHub + Claude or GPT-4o. 149 tests passing, published on PyPI. pip install autopsy-cli and you're running in 30 seconds. Looking for feedback from anyone doing on-call — what integrations should I add next? Repo: github.com/zaappy/autopsy