Connect Claude Code, Cursor, and Codex to your production stack so AI can debug logs, analyze databases, and understand real production behavior.

Hey everyone 👋 I’m Yutish, cofounder of Comie. While using AI coding tools like Claude Code and Cursor, we realized something strange: AI can write production code… but it has no idea what’s actually happening in production. It can’t see: * logs * database scale * Redis state * analytics * production errors So developers still end up deploying code just to see if something breaks. We thought that made no sense. So we built Comie.dev a production context layer for AI coding agents. You connect your production stack (Postgres, Redis, Sentry, PostHog, etc.) and now your AI can: * inspect logs * analyze production queries * debug backend issues * trace production errors directly inside Claude Code, Cursor, Codex, and more. The craziest part is how simple it is: 1. Select your tools 2. Generate tokens 3. Paste one command Setup takes under 60 seconds. We’re still very early and would genuinely love feedback from developers here. Happy to answer everything in the comments ❤️

Bridging the gap between logs and AI context is a huge pain point. Does Comie support real-time streaming of logs into an LLM context window, or is it batch-based?

This hits a real pain point — I’ve had Claude Code write what looked like solid error handling, then realized it had zero context on our actual Sentry patterns. How do you handle schema drift when the DB changes? Does the agent need explicit migration context or does it infer from query errors?

Bridging the gap between logs and AI context is a huge pain point. Does Comie support real-time streaming of logs into an LLM context window, or is it batch-based?

Comie.dev

Production context for AI with logs, DBs, and error tracking

Production context for AI with logs, DBs, and error tracking