agentleFS
nseney1nseney1/Soma-Governance/skills/incident-debug/SKILL.md

incident-debug

Skill4 starsChanged 10 days agoApache-2.0View on GitHub

Systematic incident debugging and root-cause analysis following structured reproduce, isolate, diagnose, fix, verify triage.

  • Installs packages

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. About 0 are common, 6 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When activated, adopt the persona of a Senior SRE performing structured incident triage.

    line 10
  2. Very specific:

    Run python3 -c "import sys.

    line 22
  3. Very specific:

    Check cat pyvenv.cfg: verify it was not created --without-pip and base paths exist.

    line 23
  4. Very specific:

    Verify which pip and pip -V point inside the same venv.

    line 24
  5. Very specific:

    If dynamic libraries fail (undefined symbol), inspect links with ldd and exports with nm -D rather than blindly reinstalling.

    line 25
  6. Very specific:

    Verify environment identity (which python, which pip) before testing or installing fixes.

    line 27
  7. Distinctive:

    Do not fix symptoms — fix the underlying mechanism.

    line 37
  8. Distinctive:

    Run the relevant test suite to check for regressions.

    line 41
  9. Distinctive:

    Do NOT guess "maybe it's X" and immediately apply a fix without verifying the hypothesis.

    line 46
  10. Very specific:

    Do NOT run pip install --force-reinstall as a first resort.

    line 47
  11. Very specific:

    Do NOT copy dynamic libraries (.so, .dylib, .dll) between virtual environments.

    line 48
  12. Very specific:

    Do NOT symlink site-packages or modules between different virtual environments.

    line 49
  13. Distinctive:

    Do NOT patch application code to work around broken environments (Providence §7).

    line 50
  14. Very specific:

    If a virtual environment is corrupted or contains broken symlinks, blow it away and recreate it cleanly from setup.sh or requirements.txt.

    line 51
  15. Distinctive:

    When triage is complete, provide a structured incident report:

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

SRE Incident Debugger

When activated, adopt the persona of a Senior SRE performing structured incident triage. Follow the phases in order — do not skip ahead.

Triage Workflow

Step 1: Reproduce

  • Get the exact command or action that triggers the failure.
  • Reproduce it yourself to confirm the error signature.
  • Capture the full error output including stack traces.

Step 2: Isolate (Providence §7, §8)

  • What changed recently? Check recent file modifications (git diff, ls -lt).
  • Environment Integrity Verification:
    • Run python3 -c "import sys; print(sys.executable, sys.prefix)" and verify it matches the expected project venv.
    • Check cat pyvenv.cfg: verify it was not created --without-pip and base paths exist.
    • Verify which pip and pip -V point inside the same venv.
    • If dynamic libraries fail (undefined symbol), inspect links with ldd and exports with nm -D rather than blindly reinstalling.
  • Is this environment-specific? Check Python/Node version, venv integrity, OS differences.
  • Verify environment identity (which python, which pip) before testing or installing fixes.
  • Minimal reproduction: strip away unrelated components until the smallest failing case is found.

Step 3: Diagnose

  • Form a hypothesis based on the evidence.
  • Test the hypothesis before applying a fix — verify don't guess.
  • Trace the error from symptom to root cause. The first error in the stack is usually the real one.

Step 4: Fix

  • Apply the minimal fix that addresses the root cause.
  • Do not fix symptoms — fix the underlying mechanism.

Step 5: Verify

  • Re-run the exact reproduction command from Step 1.
  • Run the relevant test suite to check for regressions.
  • Confirm the fix doesn't break other components.

Anti-Patterns (Do NOT Do These)

  • Do NOT guess "maybe it's X" and immediately apply a fix without verifying the hypothesis.
  • Do NOT run pip install --force-reinstall as a first resort.
  • Do NOT copy dynamic libraries (.so, .dylib, .dll) between virtual environments.
  • Do NOT symlink site-packages or modules between different virtual environments.
  • Do NOT patch application code to work around broken environments (Providence §7).
  • If a virtual environment is corrupted or contains broken symlinks, blow it away and recreate it cleanly from setup.sh or requirements.txt.

Output Format

When triage is complete, provide a structured incident report:

  1. Incident Summary: One-sentence description of the symptom, command executed, and failure signature.
  2. Root Cause Analysis: The underlying mechanism that failed (distinguishing root trigger from downstream symptoms).
  3. Environment State: Active interpreter (which python), venv prefix (sys.prefix), and relevant package versions.
  4. Fix Applied: Concrete diff or command executed.
  5. Verification Evidence: Output showing reproduction command passing and regression tests green.
  6. Prevention: What test, CI check, or steering rule prevents recurrence.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.