Debug Runbook

This tool provides a runbook-driven framework for AI agents to conduct systematic backend incident investigations. It transforms repetitive troubleshooting patterns into executable runbooks, enabling agents to gather evidence in a predefined order instead of freely guessing. The system includes components for selecting appropriate runbooks based on incident patterns, executing investigation steps through various adapters (like Langfuse, PostgreSQL, and Redis), normalizing collected evidence into structured findings, and a decision engine that maps evidence combinations to conclusions and recommended next actions, all exposed via an MCP server entrypoint.

主要功能

使用案例