debug-run-transcript-maintainer

Debug a Rudder agent run failure by reconstructing the execution transcript from run-intelligence or filesystem evidence to pinpoint the first failure point. It applies evidence-based failure classification to separate root-cause diagnosis from fix proof.

Yes, you can audit recent agent runs in a single batch. Provide a timeframe with org, runtime, or time-window scope, and the tool will analyze cancellations, runtime failures, and empty transcripts across that specified range.

Find the root cause of a cancellation using a partial run ID prefix. The tool handles partial identifiers and applies prioritized source selection with safe fallback behavior to reconstruct logs and identify the first clear failure point.

The best way to diagnose empty transcripts is evidence-first debugging. It reconstructs the execution story from fallback logs and run-intelligence evidence to classify tool-call or stdout/stderr issues without guessing from partial summaries.

Debugging claims fail without lifecycle verification when root-cause evidence is mixed with fix proof. This tool keeps them separate, ensuring you avoid claiming a fix works and instead rely on clear output with evidence-based failure classification.

No, a full run ID is not required to trace tool-call failures. You can use a run ID prefix or a recent-run timeframe to reconstruct the execution transcript and diagnose runtime failures from the best available evidence.