What problem does it solve?
Debugging Rudder agent run failures is hard because the surface error message often hides the real failure point inside transcript parsing, event streams, or raw execution logs.
Core Features & Use Cases
- Reconstructs the execution story: Diagnoses what happened by rebuilding a transcript from run-intelligence data, filesystem logs, and then targeted database evidence when needed.
- Handles partial run context: Supports full run IDs and run ID prefixes and helps locate likely runs when only an agent name or timeframe is available.
- Prevents common debugging missteps: Guides correct source priority and avoids incorrect assumptions like treating heartbeat events as the complete transcript.
Use case example: A run fails with a vague runtime message—use this skill to find the first clear failure point, identify whether the root cause is in transcript parsing vs runtime execution, and extract the supporting tool-call or stderr evidence.
Quick Start
Ask Rudder to debug the run by providing the run ID or prefix, for example: "Debug run ade50c99 and explain why it failed, showing the first clear failure point and the evidence."
