Back to catalog / Agent setup
Systematic debugging: find evidence before fixing causes
Reproduce, trace data flow, and inspect boundaries before choosing a repair.
Skills
Read the failure command, stack trace, recent changes, and environment differences in full, then reproduce the issue consistently. In a multi-component system, inspect each boundary’s inputs, outputs, configuration propagation, and state until you find the first point that diverges from expectation. Do not start by changing code speculatively. Form one evidence-based root-cause hypothesis and use the smallest experiment that distinguishes it from alternatives. After a fix, rerun the original reproducer and relevant regression checks. If the issue cannot be reproduced, gather more data or narrow the conditions rather than presenting a guess as the cause.
Permissions
Prerequisites
- Use for bugs, test failures, unexpected behavior, performance, or integration issues.
Permissions
- May inspect diagnostics and edit code or configuration after evidence supports a cause.
Limitations
Continue investigation when reproduction or evidence is insufficient; a coincidental one-off pass does not prove the fix.
Pinned upstream source
- Repository
- obra/superpowers
- File path
- skills/systematic-debugging/SKILL.md
- Pinned commit
bb92a77741419a4ab5f06e711a283343f1ada0c3- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: MIT
Compatible tools
Superpowers skill collection · Project test runner · Git repository tools
Main topic
Take a coding Agent change from context to verification