PR Reviews MCP Server
An MCP server that enables checking GitHub and Codeberg PR reviews, change requests, and conversations.
README
PR Reviews MCP Server
An MCP (Model Context Protocol) server that lets Cursor check GitHub and Codeberg PRs for change requests, reviews, and conversations.
Features
check_pr_reviews: Get complete review feedback, change requests, code comments, and conversations for a PRget_pr_status: Quick summary of PR review status (approvals, change requests, comment counts)
Setup
1. Install Dependencies
pip install -e .
Or with uv (faster):
uv pip install -e .
2. Create Access Tokens
For GitHub:
- Go to https://github.com/settings/tokens
- Click "Generate new token (classic)"
- Give it a name like "PR Reviews MCP"
- Select scopes:
repo(for private repositories)- Or just
public_repo(for public repositories only)
- Generate and copy the token
For Codeberg (optional):
- Go to https://codeberg.org/user/settings/applications
- Click "Generate New Token"
- Give it a name like "PR Reviews MCP"
- Select scopes:
read:repository(minimum)- Or
write:repository(if you plan to add write features later)
- Generate and copy the token
3. Configure Environment
cp .env.example .env
# Edit .env and add your tokens (at least one is required)
You need at least one token configured (GitHub or Codeberg) depending on which platforms you want to use.
4. Add to Cursor
Add this to your Cursor MCP settings (~/.cursor/mcp.json or workspace settings):
{
"mcpServers": {
"pr-reviews": {
"command": "python",
"args": ["-m", "pr_reviews_mcp.server"],
"env": {
"GITHUB_TOKEN": "your_github_token_here",
"CODEBERG_TOKEN": "your_codeberg_token_here"
}
}
}
}
Or if using uv:
{
"mcpServers": {
"pr-reviews": {
"command": "uvx",
"args": ["--from", "/Users/ryanli/Documents/GitHub/issues-mcp", "pr-reviews-mcp"],
"env": {
"GITHUB_TOKEN": "your_github_token_here",
"CODEBERG_TOKEN": "your_codeberg_token_here"
}
}
}
}
Usage in Cursor
Once configured, you can ask Cursor things like:
- "Check the PR at https://github.com/owner/repo/pull/123 for change requests"
- "What's the review status of https://codeberg.org/owner/repo/pulls/456?"
- "Show me all conversations on this PR"
The MCP server will fetch:
- ✅ All reviews (approved, changes requested, comments)
- 💬 Code-level review comments
- 📝 General PR conversation comments
- ⚠️ Change requests with reviewer feedback
See EXAMPLES.md for more detailed usage examples and sample outputs.
Tools
check_pr_reviews
Get complete PR review information including all change requests and conversations.
Input:
pr_url: Full PR URL (e.g.,https://github.com/owner/repo/pull/123orhttps://codeberg.org/owner/repo/pulls/123)
Returns:
- Change requests with reviewer feedback
- All reviews with states (approved/changes requested/commented)
- Code review comments with file paths and line numbers
- General PR conversation comments
get_pr_status
Quick summary of PR review status.
Input:
pr_url: Full PR URL (GitHub or Codeberg)
Returns:
- Number of approvals
- Number of change requests
- Comment counts
- List of reviewers requesting changes
Development
Run directly for testing:
python -m pr_reviews_mcp.server
The server communicates via stdio using the MCP protocol.
Run the test script to verify platform detection:
pip install -e .
python3 test_platforms.py
Platform Support
GitHub
- Full support for all PR reviews, comments, and conversations
- Uses PyGithub library for robust API access
- Requires
GITHUB_TOKENenvironment variable
Codeberg
- Full support via Gitea API (Codeberg runs on Gitea)
- Direct REST API calls using httpx
- Requires
CODEBERG_TOKENenvironment variable - Review state names slightly differ (
REQUEST_CHANGESvsCHANGES_REQUESTED)
Both platforms provide the same features:
- ✅ Review approvals
- ❌ Change requests
- 💬 Code-level review comments with file paths
- 📝 General PR discussion threads
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。