Token导航 LogoToken导航TokenDH.com
AI 工具操作浏览器github未标认证来源可访问clear审计未展示

local-pr-reviewer-setup本地公关审核员设置

Agent Skill

local-pr-reviewer-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

196

周安装

8

GitHub Stars

2

下载量

63
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

3

许可证

MIT

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:local-pr-reviewer-setup(本地公关审核员设置)
来源仓库:https://github.com/nartc/local-pr-reviewer
仓库路径:skills/local-pr-reviewer-setup
安装命令:
npx skills add https://github.com/nartc/local-pr-reviewer --skill local-pr-reviewer-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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/nartc/local-pr-reviewer --skill local-pr-reviewer-setup

简介

local-pr-reviewer-setup 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合整理项目状态和变更事项。

  • 适用于围绕仓库状态、代码变更或协作事项进行信息梳理的场景。
  • 通过安装命令添加到宿主环境后,可结合原始 README 核验具体用法。
  • 安装前需确认权限范围、维护状态,避免触发联网或文件读写操作。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Local PR Reviewer

A local PR review tool that lets you review code changes in a web UI and send feedback to your AI coding session.

First-time Setup

If the user hasn't set up local-pr-reviewer yet, guide them to run:

npx local-pr-reviewer setup

This will:

  • Install the review server to ~/.config/local-pr-reviewer/
  • Configure MCP tools for your coding agent
  • Optionally setup AI features for comment processing

Starting a Review Session

To start reviewing code, use the start_review_server MCP tool. This will:

  1. Start the web server if not already running
  2. Return a URL to the review interface for the current repository

Alternatively, the user can run:

npx local-pr-reviewer start

Workflow

  1. Start Review Server: Use start_review_server tool to get the review URL
  2. Open Review UI: Navigate to the URL in a browser
  3. Write Comments: Review the diff, add comments to specific lines or files
  4. Send Comments: Click "Send" to queue comments for the coding agent
  5. Receive Comments: Use check_pr_comments tool to fetch pending comments
  6. Address Comments: Make the requested changes
  7. Mark Resolved: Use mark_comment_resolved tool after addressing each comment

Proactive Comment Monitoring

When start_review_server is called or user mentions reviewing, enable continuous monitoring mode.

Monitoring Behavior

After starting PR review:

  1. Check .local-pr-reviewer-pending file in repo root after completing ANY task
  2. If pendingCount > 0, call check_pr_comments MCP tool
  3. Present new comments to user before proceeding to next task

Integration Points

Check for comments at these natural breakpoints:

  • After completing a code edit
  • After running builds/tests
  • Before responding "done" or "complete" to user
  • When idle/waiting for user input

Signal File Format

The .local-pr-reviewer-pending file contains:

{
	"sessionId": "...",
	"repoPath": "/path/to/repo",
	"pendingCount": 3,
	"updatedAt": "2024-01-15T10:30:00.000Z"
}

Quick check pattern:

  1. Read .local-pr-reviewer-pending file
  2. If pendingCount > 0, call check_pr_comments to fetch details
  3. Address comments before moving on

Example Flow

User: "Review my changes"
Agent: *calls start_review_server, gives URL to user*

User: "Fix the login bug"
Agent: *fixes bug*
Agent: *reads .local-pr-reviewer-pending, sees pendingCount: 2*
Agent: *calls check_pr_comments*
Agent: "Done with the fix. I also see you added 2 review comments - let me address those..."

Available MCP Tools

start_review_server

Starts the review web server and returns the URL for the current repository.

When to use: When user wants to review code or start a review session.

get_server_status

Check if the review server is running and get its URL.

When to use: To check server status without starting it.

check_pr_comments

Fetch pending review comments for the current repository. Comments are marked as delivered after fetching.

When to use:

  • After user has written comments in the web UI
  • When user asks to check for comments
  • Proactively while user is reviewing (poll periodically)

mark_comment_resolved

Mark a comment as resolved after addressing it. Use the comment ID from check_pr_comments.

When to use: After you've addressed a review comment. This updates the signal file with the new pending count.

list_pending_comments

List pending comments across all repositories.

When to use: To see all pending review work.

list_repo_pending_comments

List pending comments for the current repository only.

When to use: To see pending comments for the current project.

get_comment_details

Get full details of a specific comment including file path, line numbers, and content.

When to use: When you need more context about a specific comment.

Stopping the Server

To stop the running server:

npx local-pr-reviewer stop

Updating

To update to the latest version:

npx local-pr-reviewer@latest setup

Troubleshooting

Server not starting

  • Run npx local-pr-reviewer setup to ensure proper installation
  • Check if another process is using the port

Comments not appearing

  • Ensure you're in the correct repository
  • Check that the repository is registered in the review UI
  • Use check_pr_comments to manually fetch comments

MCP tools not available

  • Run npx local-pr-reviewer setup-mcp to reconfigure MCP
  • Restart your coding agent after configuration changes

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

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

Claude Code

25.54%
按下载量换算16

OpenCode

24.32%
按下载量换算15

Antigravity

16.1%
按下载量换算10

windsurf

13.52%
按下载量换算9

github-copilot

7.08%
按下载量换算4

Codex

3.13%
按下载量换算2

安全审计

暂无安全审计结果可展示。

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

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。

来源信息

继续浏览同类 Skills