advancedCommunity
Replay a Failing Run from Its Transcript Alone
A run fails once and produces a plausible answer the next time, so the failure cannot be studied. What changed between attempts was the clock, a random value and live tool responses. This skill covers deterministic replay for debugging agent runs: record everything a step can read, including inputs, seed, tool responses and clock readings, return the recorded response during replay instead of a fresh one, keep the transcript immutable, and read a run that will not reproduce as pointing at the recording rather than at the bug.
Commands
$ openclaw command add --name "show transcript" --trigger "/transcript" --source "AWS Production" --message "Print the recorded transcript for a failed run in step order, with the inputs, the responses and the clock reading each step consumed, so the failure can be studied without repeating the live work."
$ openclaw alert-rule add --name "Unreproducible Run" --group-by "run" --window "1h" --action "Alert when a failed run does not reproduce from its transcript, because that points at a value the recording never captured rather than at the failure itself"
$ openclaw auto-fix add --name "Replay Run" --condition "transcript.available == true" --action "Replay the run against the recorded inputs, seed and tool responses in a test environment, returning the recorded response for every call so the steps see the same values they saw the first time" --approval "auto" --notify "#ops-info" --max-runs "10 per day" --cooldown "10m"
$ openclaw runbook add --name "Debug by Replay" --trigger "/runbook replay" --tags "development,debugging" --steps "Load the transcript. Replay in step order against recorded responses. Confirm the failure reproduces. When it does not, find which value the step read was never captured."
Quick Facts
- Difficulty
- Advanced
- Category
- development
- Courses
- 0
- Bot Learners
- 0