systematic-debugginglisted

# Systematic debugging > Adapted from the Superpowers `systematic-debugging` skill by Jesse Vincent. It is distributed under the MIT License; see `LICENSE` in this skill directory. ## Purpose Use this process for failures, bugs, unexpected behaviour, integration problems, and performance issues. Find and explain the likely root cause before proposing a change. A workaround is not a root-cause finding. This Skill provides investigation guidance only. It does not grant remediation authority, choose tools, create a workflow, or perform a fix. Existing Run permissions and explicit user or agent authority still govern every consequential action. ## Process Complete each phase in order. Record evidence and keep facts separate from inferences. ### 1. Establish the symptom - State what failed, what was expected, and who or what was affected. - Capture the exact error, timestamps, inputs, relevant configuration, and current state. - Protect secrets and avoid copying credentials or unrelated private data into findings. ### 2. Reproduce and characterize - Reproduce the behaviour with the smallest safe case when possible. - Record whether it is deterministic, intermittent, or not reproducible. - Compare a failing case with a working case and note every meaningful difference. - If the failure is not understood, gather more evidence before suggesting a fix. ### 3. Trace the cause - Follow the data and control flow from the visible symptom toward its origin. - At each component