invoke-review

Agent Skills › AjayIrkal23/agentic-mercy-10x › invoke-review

invoke-review

GitHub

执行代码审查流程，调度santa-reviewer代理生成SANTA-REVIEW.md报告。通过运行文件夹管理上下文，分析变更文件并输出确认的发现及通过/修改请求结论，不直接修改代码。

skills/invoke-review/SKILL.md AjayIrkal23/agentic-mercy-10x

Trigger Scenarios

需要自动化代码审查时执行review act以生成审查报告时

Install

npx skills add AjayIrkal23/agentic-mercy-10x --skill invoke-review -g -y

More Options

Use without installing

npx skills use AjayIrkal23/agentic-mercy-10x@invoke-review

指定 Agent (Claude Code)

npx skills add AjayIrkal23/agentic-mercy-10x --skill invoke-review -a claude-code -g -y

安装 repo 全部 skill

npx skills add AjayIrkal23/agentic-mercy-10x --all -g -y

预览 repo 内 skill

npx skills add AjayIrkal23/agentic-mercy-10x --list

SKILL.md

Frontmatter

{
    "name": "invoke-review",
    "description": "Run the REVIEW act — dispatches santa-reviewer (default opus; opus-guard routes) and writes `SANTA-REVIEW.md` into a run folder. Single-act form of `\/invoke review -- <task>`.",
    "argument-hint": "[--run <dir>] [task]"
}

Run the review act for: $ARGUMENTS

This is /invoke review without the closers; ~/.claude/skills/invoke/SKILL.md governs anything not said here.

Run folder: --run <dir> in the arguments → reuse that folder (absolute path) and update its run.json with Edit; otherwise create a new one exactly as /invoke §2 (RUN="$(git rev-parse --show-toplevel)/.claude/runs/$(date -u +%Y%m%dT%H%M)-$SLUG", BRIEF.md, run.json with "acts": ["review"]). Never write into any other existing run folder.
Set run.json.expected_artifacts["santa-reviewer"] = "SANTA-REVIEW.md".
Agent(subagent_type="santa-reviewer", description="review: <task>", prompt=...). The prompt names $RUN/BRIEF.md, every artifact already in $RUN/, and the exact output path $RUN/SANTA-REVIEW.md. No model=: opus-guard routes it (pins, escalation, per-project mode). Artifact missing or the act failed → re-dispatch once with the prompt prefixed Previous attempt failed: <one-line reason>.
Append review to run.json.done, then report the artifact path and a one-paragraph verdict.

Review the target above, or this session's diff when it is empty: the agent reads the ACTUAL changed files, runs BREAKER + SIMPLIFIER + VERIFIER, returns only CONFIRMED findings (file:line + minimal fix) and a PASS / CHANGES REQUESTED verdict, and never edits code.