Real-time tech stack intelligence for AI agents. Tracks 10,000+ GitHub repos (5,000+ stars) across 22 categories, scoring them by popularity, growth momentum, maintenance activity, and community health. Updated daily.
Built as an MCP (Model Context Protocol) server — plug it into Claude, Cursor, or any MCP-compatible AI assistant and let it recommend the right tool for the job.
面向 AI Agent 的实时技术栈情报工具。追踪 10,000+ 个 GitHub 仓库(5,000+ stars),覆盖 22 个技术分类,从流行度、增长势头、维护活跃度和社区健康度四个维度进行评分。每日更新。
基于 MCP(模型上下文协议) 构建 — 接入 Claude、Cursor 或任何兼容 MCP 的 AI 助手,让它帮你选出最合适的开源工具。
No setup needed — use the public instance directly:
无需部署,直接使用公共服务:
Add to your MCP client config / 添加到你的 MCP 客户端配置:
{
"mcpServers": {
"agents-radar": {
"url": "http://47.253.241.19:8000/mcp"
}
}
}| Tool | Description | 说明 |
|---|---|---|
recommend_tool |
Recommend the best open-source tools for a stated need | 根据需求推荐最佳开源工具 |
get_trending |
Get trending repos by star growth velocity | 按 star 增长速度获取热门项目 |
compare_tools |
Compare tools head-to-head | 工具对比 |
check_health |
Check if a repo is actively maintained | 检查项目是否仍在活跃维护 |
list_categories |
List all tracked technology categories | 列出所有追踪的技术分类 |
- Python 3.11+
- uv (recommended / 推荐) or pip
- A GitHub personal access token (for data collection) / GitHub 个人访问令牌(用于数据采集)
git clone https://github.com/simonlin1212/agents-radar.git
cd agents-radar
# Install dependencies / 安装依赖
uv sync
# Configure environment / 配置环境变量
cp .env.example .env
# Edit .env and add your GITHUB_TOKEN / 编辑 .env 填入你的 GITHUB_TOKEN
# Create data directory / 创建数据目录
mkdir -p datauv run python -c "
import asyncio, sys
sys.path.insert(0, 'src')
from agents_radar.collector import run_collection
asyncio.run(run_collection())
"uv run agents-radarThe server starts on http://localhost:8080 with streamable-http transport.
服务启动在 http://localhost:8080,使用 streamable-http 传输协议。
docker build -t agents-radar .
docker run -p 8080:8080 -v ./data:/app/data -e GITHUB_TOKEN=ghp_xxx agents-radarAdd to claude_desktop_config.json / 添加到 claude_desktop_config.json:
{
"mcpServers": {
"agents-radar": {
"url": "http://localhost:8080/mcp"
}
}
}Deploy to your server, then point your MCP client to http://your-server:8080/mcp.
部署到你的服务器后,将 MCP 客户端指向 http://你的服务器:8080/mcp。
Authentication, AI/LLM, Databases, Ecommerce, Payments, Email, DevOps, Monitoring, CMS, Search, and more — 22 categories covering the modern tech stack. See seed_repos.json for the full list.
认证、AI/LLM、数据库、电商、支付、邮件、DevOps、监控、CMS、搜索等 — 22 个分类覆盖现代技术栈。完整列表见 seed_repos.json。
Each repo is scored (0-100) based on / 每个仓库按以下维度评分(0-100):
- Popularity / 流行度 — Stars, forks, downloads / Star 数、Fork 数、下载量
- Growth / 增长势头 — Star velocity over 7d / 30d / 90d / 7天/30天/90天的 star 增长速度
- Maintenance / 维护活跃度 — Commit recency, release frequency, issue response time / 最近提交、发版频率、issue 响应速度
- Community / 社区健康度 — Contributors, open issues ratio, documentation quality / 贡献者数量、open issue 比例、文档质量
MIT