agentleFS
itsofficialadiitsofficialadi/boundiff/.agents/skills/root-cause-investigation/SKILL.md

root-cause-investigation

Skill0 starsChanged 26 days agoMITView on GitHub

Investigate intermittent bugs, regressions, errors, or failing CI to isolate a defensible root cause before changing code. Diagnosis-only; use bug-fix-verification for remediation.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not jump from the first error message to a patch.

    line 8
  2. Distinctive:

    Run small, controlled experiments when the environment permits.

    line 13
  3. Distinctive:

    Update hypotheses as evidence arrives.

    line 14

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

Root-cause investigation

Do not jump from the first error message to a patch. This skill investigates; it does not remediate.

  1. Establish symptoms with timestamps, affected versions, inputs, expected behavior, and the earliest reliable failure signal. Separate the user's report from observed evidence.
  2. Form no more than three competing explanations. State what observation would discriminate between them.
  3. Trace the relevant path: caller, state transitions, data shape, exceptions, retries, configuration, and recent changes. Search narrowly before searching everywhere.
  4. Run small, controlled experiments when the environment permits. Start with deterministic local reproductions; avoid writing to production or external systems. Record command, input, and outcome.
  5. Update hypotheses as evidence arrives. Do not label correlation as root cause. If no explanation survives testing, explicitly report a bounded unknown.
  6. Hand off the smallest candidate correction and a proposed regression check to $bug-fix-verification. Do not edit the reviewed code during diagnosis.

Special cases

  • CI-only failure: compare runtime versions, environment, time, ordering, data, and resource constraints rather than immediately retrying.
  • Flaky test: check shared state, parallelism, nondeterministic order, leaked timers, and dependencies before increasing timeouts.
  • Incident: treat secrets, live customer data, and deployment actions as separate authorization boundaries.

Output

Symptom, evidence timeline, hypotheses tested, most likely cause with confidence, reproduction, next safe action. Say clearly when the cause remains unconfirmed.

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.