mohammadhprp/teammate/investigate-issue
Claude Code and agent skill by mohammadhprp/teammate
- community
Checking that the SKILL.md file is still in the repository.
Investigate a question about a codebase read-only: gather evidence from files, logs, history, and tests, separate a proven conclusion from a hypothesis, change nothing unrelated, and report the evidence with a confidence level. Use when a worker is delegated a diagnosis, root-cause, or feasibility question that does not include a fix — even when the wording is 'figure out why' and the cause looks obvious enough to fix on the spot.
Install
npx vskill@latest install mohammadhprp/teammate/investigate-issueSKILL.md
The same SKILL.md was published earlier as mohammadhprp/teammate/investigate-issue.
Investigate issue
An investigation earns trust by making its answer checkable, not by sounding certain. The requester should be able to follow your evidence to your conclusion and tell which parts are proven and which are still guesses.
This is read-only. If the assignment wants a fix, that is implement-task, and the change must be verified separately; mixing the two contaminates the evidence with the change you wanted to make. When you find the cause, stop and report it.
When to use
- You are asked a question about the codebase: why something fails, how a
behavior works, whether an approach is feasible, or where a value comes from.
- You are asked to diagnose something without a fix assignment.
When not to use
- The assignment includes a fix — that is
implement-task, then
verify-change.
- The behavior is already understood and the fix is trivial — just make it.
- You are checking a completed change — that is
review-task/review-change.
Inputs
- The question, stated precisely enough to answer.
- The project root and context (see
load-project-context). - The relevant surface: source, tests, history, logs, configuration.
Procedure
- Restate the question as something testable. "Why does X fail?" becomes "X
fails when Y; is the cause Z?" A testable question keeps the investigation from drifting into unrelated reading.
- Gather evidence, cheapest first. Read the relevant code and tests; check
history with git log / git blame for when behavior changed; reproduce the failure or run the test that exhibits it. Reproduce before theorizing whenever you can.
- Test the hypothesis against the evidence. Follow the code and the run, not
plausibility. verify-evidence applies here: a conclusion needs a check behind it, not a story.
- Separate conclusion from hypothesis. Label what the evidence proves and
what remains unconfirmed, naming the check that would confirm it.
- Change nothing unrelated. Read-only: no drive-by edits, no fixes, no
cleanup. If you must run something, leave the tree as you found it.
- Report evidence plus confidence. Cite file and line, the command and its
output, your confidence, and what would move it.
Output
The question, the evidence, the conclusion, the confidence, and the open questions — in the handoff-report shape, through report-result.
Failure and escalation
- Cannot conclude → report the evidence gathered and the open questions rather
than dressing a hypothesis as a finding; use raise-blocker if a decision is needed to continue.
- The investigation points to a fix that is in scope but unassigned → recommend
it; do not perform it.
- The question cannot be answered as written → ask one clarifying question with
raise-blocker instead of investigating the wrong thing.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingsrc/skills/investigate-issue/SKILL.md
Popularity
- Repository stars
- 2
- Repository forks
- 1
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/mohammadhprp/teammate/investigate-issue)