Codag addresses the challenge of handling large-scale system logs for AI agents by providing log compression specifically designed for agent workflows. The tool reduces the number of log tokens required by up to 95% while preserving the debugging signals necessary for effective troubleshooting. Codag identifies and extracts only the lines that matter from raw logs, grouping them into ranked patterns with counts and citing real line numbers, along with relevant context. This approach ensures that nothing is summarized away or invented, and the most suspect patterns are prioritized for review.
The platform is suitable for both individuals and organizations managing production environments with extensive logging, such as infrastructure and platform teams. Codag is designed to work with millions of log lines per day, making it practical for scenarios where the volume of logs exceeds what AI agents can process in a single context window. Output from Codag is compact plain text that can be pasted into AI tools like Claude, GPT, or MCP-based workflows, allowing agents to process only the essential information at a fraction of the token cost.
Codag integrates with a variety of log sources and platforms, including AWS, Vercel, Railway, Kubernetes, Docker, Datadog, and Sentry, among others. It can wrap any existing log command, providing drop-in compression over HTTPS. Setup involves installing a hook and MCP server, after which AI agents like Claude or Codex can read logs through Codag without additional configuration. The tool offers both a free deterministic version and a Pro version that uses inference-based compaction.
Codag is available as an open source solution, with a free tier that is stated to be available forever. Installation can be performed via a shell script, and the service is positioned as a drop-in enhancement for existing agent log workflows. Its primary class is systems log compression for AI agents, focusing on retaining actionable evidence from large log sets while minimizing token usage.
Codag is an Infrastructure & Backend product. It helps developers visualize and debug complex LLM workflows and API calls in their codebase. Codag is an open-source project aimed at developers. The project is open source (MIT). It runs on the web.
Behind Codag is michaelzixizhou, and the product first shipped in 2025. The project is developed in the open on GitHub with 540 stars. Across PulseGate's embedding index, Codag has few near neighbours, marking it as relatively distinct. Among its 4 catalogued features are LLM workflow visualization, code analysis, and interactive graphs.
Latest indexed changes and source events
Other apps tracked under the same category.
