agentleFS
WrosinskiWrosinski/heddle/.claude/skills/root-cause-analysis/SKILL.md

root-cause-analysis

Skill0 starsChanged 57 days agoMITView on GitHub

Root-cause a bug whose symptom may mislead: intermittent, flaky, or already "fixed" once and back. Produces the causal chain with evidence and a solution proposal, not the implementation.

18 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use observability, adversarial-review and 6 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • OpenAISkill

    observability

    Vercel Observability expert guidance — Drains (logs, traces, speed insights, web analytics), Web Analytics, Speed Insights, runtime logs, custom events, OpenTelemetry integration, and monitoring dashboards. Use when instrumenting, debugging, or optimizing application performance and user experience on Vercel.

    Also in 3 projects that use this

  • MicrosoftSkill

    adversarial-review

    Adversarial review of a system design from 6 critical perspectives -- SRE, security, staff engineer, finance, operator, and developer advocate. Produces a unified risk assessment. Use for INTERACTIVE on-demand reviews during a design conversation (/adversarial-review). For RECIPE-DRIVEN reviews (where prior step context is needed), use the systems-design-critic agent instead.

    Also in 3 projects that use this

  • GoogleSkill

    domain-modeling

    Build and sharpen a project's domain model. Use when the user wants to pin down domain terminology or a ubiquitous language, record an architectural decision, or when another skill needs to maintain the domain model.

    Also in 4 projects that use this

  • GoogleSkill

    improve-codebase-architecture

    Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

    Also in 3 projects that use this

  • MicrosoftSkill

    handoff

    Compact the current conversation into a handoff document for another agent to pick up.

    Also in 3 projects that use this

  • +3 more from google

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 17 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use heddle/resources/decision-routing.md for ownership and question quality.

    line 18
  2. Distinctive:

    Use the governing spec and applicable decisions for expected behavior.

    line 26
  3. Distinctive:

    Do not invent a reproduction or treat an old successful run as current evidence.

    line 36
  4. Distinctive:

    Follow the failing execution and state transitions far enough to identify the condition that produces the symptom.

    line 40
  5. Distinctive:

    Use causal tracing, fault trees, or difference analysis as appropriate.

    line 42
  6. Distinctive:

    Do not force an unchangeable upstream outage into a fictional locally fixable cause.

    line 53
  7. Distinctive:

    Use the requested report format.

    line 96

“Common” is measured: kept by at least 6 in 10 of the 17 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 Analysis

Explain the reported failure through a supported causal mechanism and propose the change that would remove it. Distinguish a verified mechanism from proof that it caused this particular incident. If the incident cannot be established, report the remaining uncertainty and the evidence needed to resolve it.

Decision ownership

Use heddle/resources/decision-routing.md for ownership and question quality; feature kickoff delivers the resolved host policy in full. Preserve this skill's analysis/authoring boundary and applicable standing authorization. A review or diagnosis request alone grants no implementation authority.

Scope and Authority

Investigate the reported symptom, relevant failure paths, and plausible causes. Use the governing spec and applicable decisions for expected behavior; use source, tests, logs, and reproducible observations for actual behavior. Read-only inspection and relevant hermetic diagnostic checks are in scope. A diagnosis request alone does not authorize implementation, permanent tests, live/paid reproduction, production mutation, or unrelated cleanup. Honor additional authorization already supplied for the task.

State expected versus actual behavior, affected inputs/environment, timing, scope, and reproducibility from available evidence. Ask for missing information only when it changes the investigation; finish independent checks while a question is open. Do not invent a reproduction or treat an old successful run as current evidence.

Causal Evidence

Follow the failing execution and state transitions far enough to identify the condition that produces the symptom. Compare working and failing cases when useful. Use causal tracing, fault trees, or difference analysis as appropriate; there is no required number of whys, branches, or hypotheses.

Evaluate plausible alternatives against both supporting and contrary evidence. Record an untested alternative as untested, not disproven. An absence of contrary evidence is not itself confirmation. Correlated timing identifies a lead; the mechanism and observations establish causation.

A supported root cause explains the observed manifestation conditions and identifies the actionable defect, violated contract, or external condition. Explain how the proposal removes the mechanism or prevents its recurrence under those conditions. Do not force an unchangeable upstream outage into a fictional locally fixable cause. Distinguish the cause from a mitigation, detection improvement, or recovery measure. Reverting a fix can restore a symptom for both a causal fix and a workaround, so reversion alone does not distinguish them.

For example, queue delay is a symptom when a trace shows another worker holding a shared connection throughout a large transaction. Reducing transaction scope is a causal proposal only if the transaction's atomicity is unnecessary under the actual contract. Priority scheduling may mitigate delay without removing connection starvation. Both proposals need their own evidence and trade-offs.

Reuse applicable investigation evidence. Recheck it when a relevant source, environment, input, or contradictory observation changes. Distinguish inspection, reproduction, and inference; a passing suite does not disprove an untested incident.

Named traps

  • Premature termination: stopping at a plausible label without explaining the mechanism or the observed conditions. Trace the decision-relevant links; mark an inaccessible link uncertain instead of extending the chain by guesswork.
  • Single-thread pursuit: accepting an early hypothesis without testing plausible alternatives. State what observations distinguish them and which checks were possible. Missing contrary evidence is neither confirmation nor proof that contrary evidence must exist.
  • Symptom-fix conflation: presenting mitigation, detection, or recovery as removal of the cause. Explain which mechanism the proposal changes and under what conditions recurrence is prevented. An external cause can be established even when only a local mitigation is available.

Delegation

Delegate distinct hypotheses or independent path traces when that adds evidence or saves time while the caller continues useful work. Intermittency or a prior failed fix alone does not require parallel agents. Follow docs/workflow/subagent-exploration-patterns.md for bounded briefs, model tier, current tool capabilities, and evidence synthesis. A path trace is exploration work; an investigation whose product is weighing contradictory evidence or adjudicating a prior failed fix is the named case for the top tier. Give each investigator the same observed problem and require supporting evidence, contrary evidence, and unresolved limits.

Deliverable and Completion

Use the requested report format; otherwise provide:

  • Finding: confirmed cause or leading explanation, confidence, and incident-specific uncertainty, followed by expected/actual behavior and conditions.
  • Evidence: causal chain with source locations and observations, hypotheses tested, contrary evidence addressed, and relevant checks with their actual results.
  • Proposal: recommended change and its causal connection, alternatives when materially different, affected contracts, regression risks, and required authority.
  • Verification plan: the smallest meaningful check that distinguishes the diagnosis from its alternatives and proves the proposed behavior if implemented.

The investigation is complete when the evidence supports the cause and proposal, or when available in-scope checks are exhausted and the remaining evidence gap is explicit. An unsupported causal link stays uncertain. End with the diagnosis and next decision; do not implement the fix unless the user authorizes it.

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.