AI Diff Review MCP
Enables AI agents to edit files in VS Code with an interactive diff review panel, allowing users to accept or reject changes.
README
AI Diff Review (MCP)
A Model Context Protocol (MCP) server that runs directly inside your Visual Studio Code. It empowers AI agents with advanced, state-of-the-art file editing capabilities while ensuring you stay in complete control of your codebase.
Instead of AI agents blindly overwriting files, this extension generates an interactive diff review panel inside VS Code, allowing you to explicitly Accept or Reject any changes proposed by the AI.
Features
- Interactive Diff Reviewer: All AI-proposed edits are presented in a native VS Code split-screen diff view. You can review the changes before finalizing them.
- Surgical Code Edits (
edit_block): The AI can search for a specific block of code and replace it. - Fuzzy Search Fallback: If the exact string isn't found, the powerful Levenshtein fuzzy matching engine will find the closest match and guide the AI to correct its prompt automatically.
- Live Diagnostics Feedback: Once an edit is proposed, the extension automatically hooks into VS Code's language servers (TypeScript, ESLint, Python, etc.) and reports any syntax errors or warnings back to the AI immediately!
- State Management: Multiple edits to the same file are intelligently combined into a single pending diff.
Installation
You can either install the pre-compiled VSIX or build from source.
Option 1: Install from VSIX (Recommended)
- Go to the Actions tab.
- Download the latest
ai-diff-review-mcp-vsixartifact. - Open VS Code, press
Ctrl+Shift+P(orCmd+Shift+Pon Mac). - Type and select Extensions: Install from VSIX...
- Select the
.vsixfile you just downloaded. - Reload VS Code.
Option 2: Build from Source
- Clone the repository:
git clone https://github.com/iiNothh/ai-diff-review-mcp.git - Run
npm installto install dependencies. - Run
npm run compileto build the typescript files. - Run
npx @vscode/vsce packageto create the VSIX file. - Install as mentioned in Option 1.
Usage / Connecting the AI Agent
Once the extension is installed and VS Code is running, the MCP server automatically starts in the background and listens for SSE connections on port 6070.
To connect your AI client, add the appropriate configuration to your MCP settings file:
For Antigravity (mcp_config.json):
{
"mcpServers": {
"ai-diff-review-mcp": {
"serverUrl": "http://127.0.0.1:6070/mcp"
}
}
}
For other standard MCP Clients (mcp.json):
{
"servers": {
"ai-diff-review-mcp": {
"url": "http://127.0.0.1:6070/mcp"
}
}
}
The AI agent will then be equipped with the powerful write_file and edit_block tools seamlessly connected to your VS Code session.
Contributing
Feel free to open Issues or Pull Requests if you find a bug or want to enhance the file-edit engine further.
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。