Token导航 LogoToken导航TokenDH.com
开发敏感数据github未标认证来源可访问许可证需确认审计提醒

repo-ai-setup回购 AI 设置

Agent Skill

repo-ai-setup 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

225

周安装

9

GitHub Stars

12

下载量

73
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:repo-ai-setup(回购 AI 设置)
来源仓库:https://github.com/walletconnect/skills
仓库路径:skills/repo-ai-setup
安装命令:
npx skills add https://github.com/walletconnect/skills --skill repo-ai-setup
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/walletconnect/skills --skill repo-ai-setup

简介

用于处理 GitHub 仓库、Issue 和 Pull Request 信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中整理代码变更和协作事项。
  • 可结合来源仓库和原始 README 核验具体用法。
  • 安装前建议确认权限范围、维护状态及是否会触发联网或文件读写。
  • 安装方式:通过 npx skills add 从 GitHub 仓库安装。

SKILL.md

Repo AI Setup

Set up AI agent documentation and automated PR review infrastructure for a WalletConnect repository.

Arguments: $ARGUMENTS

Argument Handling

  • $ARGUMENTS contains "just docs" or "skip workflow" → run Phases 0–2 only
  • $ARGUMENTS contains "just workflow" or "just auto-review" → run Phases 0 + 3 only
  • No arguments → all phases

Phase 0 — State Detection

Detect the current repo state before doing anything:

  1. Check if AGENTS.md exists in repo root
  2. Check if CLAUDE.md exists — if so, determine: regular file or symlink? If symlink, where does it point?
  3. Check if .github/workflows/claude-review.yml or .github/workflows/claude-auto-review.yml exists
  4. Detect default branch: gh repo view --json defaultBranchRef -q.defaultBranchRef.name 2>/dev/null \ || git symbolic-ref refs/remotes/origin/HEAD 2>/dev/null | sed 's@.*/@@' \ || echo "main"
  5. Verify this is a git repo (git rev-parse --git-dir)

Report state summary to user. Example:

Repository state:
  AGENTS.md:      missing
  CLAUDE.md:      exists (regular file)
  Auto-review:    missing
  Default branch: main
  Git repo:       yes

Skip any phases whose artifacts already exist.


Phase 1 — Create AGENTS.md

Three branches based on detected state:

Case A: AGENTS.md already exists

Skip. Report: "AGENTS.md already exists — skipping."

Case B: CLAUDE.md exists as regular file (not symlink)

Convert it to an agent-agnostic AGENTS.md:

  1. cp CLAUDE.md AGENTS.md
  2. Edit AGENTS.md:

- Replace # CLAUDE.md header with # AGENTS.md (if present) - Replace "This file provides guidance to Claude Code" or similar → "This file provides guidance to AI coding agents" - Replace "Claude Code" → "AI coding agent" and "Claude" → "AI agent" in descriptive text - Preserve all technical instructions, commands, code blocks, and configuration details unchanged - Preserve any references to .claude/ paths, CLAUDE.md symlinks, or Claude-specific tooling config (these are real paths/tools)

  1. Delete the original CLAUDE.md (it will be recreated as a symlink in Phase 2)

Case C: No agent docs exist (neither AGENTS.md nor CLAUDE.md)

Generate agent docs using Claude Code's /init command via a subagent:

  1. Launch a Task agent with subagent_type: general-purpose:

- Instruct it to run Claude Code's /init command as a CLI subprocess: claude -p "/init" --allowedTools "Bash,Read,Write,Edit,Glob,Grep" - This generates CLAUDE.md in the repo root with real codebase analysis - If claude CLI is not available, the subagent should fall back to manual codebase analysis: scan the repo for build files, test configs, CI workflows, source structure, READMEs, etc., and write a comprehensive AGENTS.md directly (use the format guide in REFERENCE.md)

  1. After the subagent completes:

- If CLAUDE.md was generated, rename it: mv CLAUDE.md AGENTS.md - Apply the same agent-agnostic edits as Case B step 2 - If the subagent wrote AGENTS.md directly (fallback path), verify it exists


Phase 2 — Create CLAUDE.md Symlink

Ensure CLAUDE.md is a symlink pointing to AGENTS.md:

  1. If CLAUDE.md is already a symlink to AGENTS.md → skip
  2. If CLAUDE.md is a symlink to something other than AGENTS.mdwarn the user and ask how to proceed (do not overwrite)
  3. If CLAUDE.md exists as a regular file at this point → warn (shouldn't happen after Phase 1), ask user
  4. Otherwise, create the symlink: ln -s AGENTS.md CLAUDE.md
  5. Verify: readlink CLAUDE.md # should output "AGENTS.md"

Phase 3 — Setup Claude Auto-Review Workflow

Set up the GitHub Action for automated Claude PR reviews.

  1. If .github/workflows/claude-review.yml already exists → skip
  2. Also check for .github/workflows/claude-auto-review.yml → if exists, skip (alternate naming)
  3. Create .github/workflows/ directory if it doesn't exist: mkdir -p.github/workflows
  4. Read the workflow template from REFERENCE.md (Section: Claude Auto-Review Workflow Template)
  5. Replace {DEFAULT_BRANCH} with the actual default branch detected in Phase 0
  6. Write the file to .github/workflows/claude-review.yml
  7. Remind the user: they need to add ANTHROPIC_API_KEY as a repository secret:

- Go to Settings → Secrets and variables → Actions → New repository secret - Name: ANTHROPIC_API_KEY - Value: their Anthropic API key

  1. Ask the user if they want to add a project_context input to the workflow for project-specific review guidance. If yes, ask them for the context string and add it to the workflow per the REFERENCE.md example.

Not a git repo?

If Phase 0 detected this is not a git repo, skip Phase 3 entirely. Still create AGENTS.md + symlink.


Phase 4 — Summary

Report what was created, modified, and skipped:

Setup complete:
  ✓ Created AGENTS.md (from /init analysis)
  ✓ Created CLAUDE.md → AGENTS.md symlink
  ✓ Created .github/workflows/claude-review.yml
  ⚠ Remember to add ANTHROPIC_API_KEY repo secret

Suggested commit message:
  feat: add AI agent documentation and auto-review workflow

Do NOT auto-commit. Let the user review and commit.


Edge Cases

ScenarioHandling
CLAUDE.md symlinks to non-AGENTS.md targetWarn user, ask before changing
Not a git repoSkip Phase 3, still create docs + symlink
gh CLI not availableFall back to git commands for branch detection
claude CLI not available in subagentSubagent does manual codebase analysis
AGENTS.md + CLAUDE.md both exist as regular filesWarn about duplication, ask user which to keep
.github/workflows/ has a differently-named Claude review workflowCheck for both claude-review.yml and claude-auto-review.yml

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

31.87%
按下载量换算23

Claude

31.66%
按下载量换算23

Cursor

20.23%
按下载量换算15

Gemini CLI

9.94%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

可疑

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills