Python debug
My personal pi harness configuration.
npx -y skills add nyquistwilder/personal-pi --skill python-debugAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Python debugging workflow for reproducing failures, tracing exceptions, isolating root cause, and fixing defects with focused regression tests in uv-managed src/tests projects.
SKILL.md
2.8 KB, 511 tokens by cl100k_base, as published. Nobody here has run it
Python Debug
Rule
Debug from evidence. Reproduce the failure, isolate the root cause, make the smallest correct fix, and add a regression test when practical.
Do not make speculative changes unrelated to the observed failure.
Hard Stops
Stop and ask before proceeding when:
- The failure cannot be reproduced and there is no clear expected behavior.
- Reproduction requires production systems, live credentials, mutable user data, or destructive operations.
- Logs, traces, crash dumps, or fixtures may contain secrets or sensitive data.
- The likely fix changes public API, persistence schema, CLI/API contracts, or security boundaries beyond the reported defect.
- The task is primarily performance tuning, security incident response, or test-suite design; use the more specific skill.
Workflow
- Capture the exact failing command, traceback, input, environment, and expected behavior.
- Inspect recent changes, relevant tests, configuration, lockfile, and task wrappers.
- Reproduce with the narrowest command, preferably through uv.
- Reduce the failure to the smallest failing test, fixture, or script.
- Identify root cause before editing. Distinguish symptom, trigger, and underlying defect.
- Add or update a regression test when practical and safe.
- Implement a targeted fix.
- Run the narrow regression test, then related tests, then
just checkor the repository validation command.
Debugging Practices
- Prefer deterministic tests over print-driven debugging.
- Use structured logs,
caplog, or temporary local instrumentation only when useful; remove temporary instrumentation before completion. - Check import paths, editable install state, uv groups, environment variables, timezones, filesystem assumptions, and network/database boundaries.
- For async failures, inspect cancellation, timeout, event-loop ownership, blocking calls, and resource cleanup.
- For HTTP failures, capture method, URL shape, timeout, status handling, retry behavior, and response parsing without logging secrets.
- For data failures, preserve a minimal non-sensitive fixture that demonstrates malformed, missing, or edge-case input.
Review Guidance
A good bug fix should explain why the failure happened, why the patch fixes that cause, and how tests prevent recurrence. Flag broad rewrites, swallowed exceptions, missing regression coverage, and fixes that only satisfy one incidental traceback.
Completion
Report reproduction steps, root cause, fix summary, regression coverage, commands run, and any unverified assumptions or remaining risks.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.