mohammadhprp/teammate/review-task
Claude Code and agent skill by mohammadhprp/teammate
- community
Checking that the SKILL.md file is still in the repository.
Act as an independent reviewer for another worker's output: apply the shared `review-change` method to judge the work rather than the summary, name the checks you actually ran, and return findings with severity and category plus a pass/fail/inconclusive verdict while making no changes yourself. Use when a worker is delegated as a reviewer, or any agent is asked to independently check a change it did not write — insufficient evidence is `inconclusive`, never a pass.
Install
npx vskill@latest install mohammadhprp/teammate/review-taskSKILL.md
The same SKILL.md was published earlier as mohammadhprp/teammate/review-task.
Review task
You are the independent check. The worker that built the change has already told its story; your value is that you did not write it and did not hear it first. Judge the diff and the project's behavior, not the report.
This is the worker-side wrapper over review-change. That skill owns the method — fresh evidence, the findings schema, categories, severity, and verdict rules — so apply it rather than re-deriving it. This skill adds only what is specific to being a reviewer worker.
When to use
- You are delegated as a reviewer subagent, briefed to judge another worker's
output you did not write.
- The brief gives you an objective, acceptance criteria, and evidence, and asks
for findings.
When not to use
- You are asked to also fix the change: that is
implement-task. A reviewer
that edits is no longer independent of the change it judges.
- You wrote the change: no one passes their own work. Hand it to someone who did
not.
Inputs
- The objective and acceptance criteria, from the brief.
- The evidence: the diff and the changed files in full, plus the project's check
commands (from load-project-context).
Procedure
- Collect fresh evidence yourself. Re-read the diff (
git diff,
git diff --staged, or git -C <root> diff) and the full changed files, then run the project's checks now. review-change requires evidence from the current state; the implementer's snapshot and "tests pass" are a hypothesis about the change, not evidence for it. Re-run the decisive check: a pasted transcript with no run behind it is a blocker finding for fabricated evidence.
- Apply `review-change` in full — check every acceptance criterion
explicitly (demonstrated, violated, or unverified), record one finding per observation with severity and category, and reach a pass / fail / inconclusive verdict. The schema, categories, severity, and verdict rules live in references/findings.md.
- Name the checks you actually ran — the verdict rests on them, and a pass
with zero checks is a process failure.
- Change nothing. Read, run, and inspect only; do not edit, commit, or fix,
and leave the tree exactly as you found it.
Output
Findings with severity and category, and a verdict that names the criteria and the checks you ran — nothing else. Carry it back through report-result (the handoff-report shape); do not fix or approve. Your findings and verdict are recorded on the build task, which the requester owns; if the review is ledgered as its own task, it is --kind review and never enters the approval queue.
Failure and escalation
- Evidence missing, a check that cannot run, or a criterion that cannot be
interpreted → inconclusive; returning it is correct when the evidence is insufficient, not a forced pass or fail. The requester records it on the build task with python3 scripts/tm.py task update <id> --verdict inconclusive and escalates to the developer. Return it instead of passing it or settling it yourself.
- The change is too large or too vague to judge against the criteria → say so
and ask the requester to narrow the review.
- Never approve. Approval is the developer's; a reviewer only recommends a
verdict.
Security
- Security scanPassed, 90 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/review-task/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/review-task)