Diagnose Bug
Evidence-led, read-only bug diagnosis using minimal reproduction, boundary isolation, ranked hypotheses, and explicit falsification tests before any fix is proposed.
Install
npx skills add flurdy/agent-skills --skill diagnose-buggit clone --depth 1 https://github.com/flurdy/agent-skills.git /tmp/agent-skills
mkdir -p ~/.claude/skills
cp -r /tmp/agent-skills/skills/diagnose-bug ~/.claude/skills/diagnose-bugPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Diagnose Bug Find what causes an observed failure before proposing implementation. Work from evidence: ```text reproduce → isolate → hypothesize → falsify → conclude ``` Diagnosis is read-only. It may run safe tests and inspect code, logs, configuration shape, and tracker context, but it does not edit code, tests, configuration, or trackers. ## When to use Use `/diagnose-bug` when: - the symptom is reproducible but the cause is unknown; - a regression crosses modules, services, environments, or asynchronous boundaries; - the failure is intermittent, input-specific, timing-sensitive, or production-only; - several plausible explanations exist and speculative fixes would be risky; - the user asks for root-cause analysis, investigation, isolation, or diagnosis. Usually skip it when: - a deterministic failing test and direct source trace already establish the cause; - the root cause is known and the request is purely to implement the fix; - the work is a feature, refactor, architecture plan, or post-fix verification; - the only question is whether an existing implementation satisfies requirements—use `/verify-task` instead. ## Usage ```text /diagnose-bug <symptom, bead id, Jira key, log excerpt, or failing command> /diagnose-bug ABC-123 /diagnose-bug skills-123 /diagnose-bug --no-prompt <bug> ``` ## Operating boundary - Do not edit source, tests, snapshots, fixtures, lockfiles, generated files, configuration, or trackers during diagnosis.
Details
- Skill name
diagnose-bug- Repository
- flurdy/agent-skills
- Path
skills/diagnose-bug/SKILL.md- License
- MIT
- Repo updated
- 17 November 2026
FAQ
What does the Diagnose Bug skill do?
Evidence-led, read-only bug diagnosis using minimal reproduction, boundary isolation, ranked hypotheses, and explicit falsification tests before any fix is proposed.
How do I install the Diagnose Bug skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add flurdy/agent-skills --skill diagnose-bug. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Diagnose Bug popular?
It lives in flurdy/agent-skills, which has 9 GitHub stars and 0 forks.
Related skills
Architect
9flurdy/agent-skills
Architecture and implementation planning gate for complex or high-blast-radius work. Produces evidence-backed decisions, reviewable slices, acceptance evidence, and conditional human review ownership without editing code.
Artifact Hygiene
9flurdy/agent-skills
Run a local-only, read-only advisory audit of publishable working-tree files and unpublished branch history with redaction-safe findings.
Backlog Groom
9flurdy/agent-skills
Auto-detects workspace unfinished backlogs across unique stores; otherwise local open beads. Flags thin prose, missing labels, drift, lifecycle and duplicates. Report-only; mutations need approval, destructive ones individually.
Pedantic Review
9flurdy/agent-skills
Read-only craft review of changed code and test design: reuse, placement, complexity, and repository conventions. Requirements, coverage sufficiency, execution, and fixes have separate owners.
Testing Setup
9PanPanFR/oh-my-openkilo
Analyze and create a testing strategy for Android apps - install testing libraries, set up test infrastructure, create harnesses for unit tests, UI tests, screenshot tests, and end-to-end tests.
Skills Review
9flurdy/agent-skills
Explicit, skeptical review of a source skill collection. Screens the catalog, tests sampled skills' necessity, and proposes evidence-backed trims, merges, retirements or fixes; read-only until selected findings go to triage.
Browse the most popular Claude skills, all testing & qa skills, or write your own with the skills generator.