GeM MCP Server
Enables users to search, track, and analyze procurement data from India's Government e-Marketplace (GeM) for live and stored bid listings. It provides specialized tools for monitoring drone-related procurement opportunities and generating detailed bid analysis reports.
README
GeM MCP Server — Government e-Marketplace Intelligence
A fully-featured, Claude-compatible Model Context Protocol (MCP) server for interacting with India's Government e-Marketplace (GeM) bid data.
Deploy on Render and connect from Claude Code, Claude Web, or any MCP-compatible agent.
Features
🔧 MCP Tools (9 tools)
| Tool | Description |
|---|---|
search_bids |
Search the live GeM portal by keyword, date, ministry, org |
get_latest_bids |
Crawl latest N pages of bid listings |
search_drone_bids |
Find all drone/UAV procurement opportunities |
get_stored_bids |
Query stored bids with filters (status, ministry, type, etc.) |
get_bid_detail |
Fetch extended info from a bid detail page |
lookup_bid |
Look up a specific bid by number from the database |
get_bid_stats |
Database statistics (counts, top ministries, etc.) |
cleanup_old_bids |
Delete bids older than N days |
export_bids |
Export all stored bids as JSON |
📄 MCP Resources (4 resources)
| URI | Description |
|---|---|
gem://keywords |
Monitored drone/UAV keywords list |
gem://config |
Server configuration (non-sensitive) |
gem://stats |
Live database statistics |
gem://urls |
GeM portal URL structure and endpoints |
💬 MCP Prompts (3 prompts)
| Prompt | Description |
|---|---|
analyze_procurement |
Sector-specific procurement analysis workflow |
drone_opportunity_report |
Comprehensive drone/UAV opportunity report |
bid_deep_dive |
Detailed analysis of a specific bid |
Quick Start
1. Clone and Install
git clone <your-repo-url>
cd gem_mcp
pip install -r requirements.txt
2. Configure MongoDB
Copy .env.example to .env and set your MongoDB connection string:
cp .env.example .env
# MongoDB Atlas free tier
MONGODB_URI=mongodb+srv://user:password@cluster0.xxxxx.mongodb.net/?retryWrites=true&w=majority
MONGODB_DB=gem_mcp
3. Run Locally
python server.py
Server starts at http://localhost:8000:
- MCP endpoint:
http://localhost:8000/mcp - Health check:
http://localhost:8000/health
4. Connect a Client
Claude Code:
claude mcp add --transport http gem-server http://localhost:8000/mcp
MCP Inspector:
npx -y @modelcontextprotocol/inspector
# Connect to: http://localhost:8000/mcp
Deploy to Render
Option A: Blueprint (one-click)
- Push this repo to GitHub
- Go to Render Dashboard → New → Blueprint
- Connect your repo — Render reads
render.yamlautomatically - Set the
MONGODB_URIenvironment variable in the Render dashboard
Option B: Manual Docker Deploy
- New → Web Service → Docker
- Point to your repo
- Set environment variables:
MONGODB_URI= your Atlas connection stringMONGODB_DB=gem_mcp
- Health check path:
/health
After Deploy
# Add to Claude Code
claude mcp add --transport http gem-server https://your-app.onrender.com/mcp
Architecture
Client (Claude Code / Claude Web / Any MCP Agent)
│
│ MCP Protocol (Streamable HTTP)
▼
┌─────────────────────────────────┐
│ server.py (FastMCP + Starlette)│ ← Render Web Service
│ POST /mcp — MCP endpoint │
│ GET /health — health check │
├─────────────────────────────────┤
│ gem_scraper.py │ ← Scraping + parsing logic
│ (requests + BeautifulSoup) │
└─────────┬───────────────────────┘
│ │
▼ ▼
MongoDB Atlas bidplus.gem.gov.in
(persistent (live scraping)
storage)
Data Fields
| Field | Description |
|---|---|
bid_no |
Unique bid number (e.g. GEM/2024/B/123456) |
title |
Bid title / product description |
quantity |
Number of units required |
uom |
Unit of measurement |
start_date |
Bid opening date |
end_date |
Bid closing deadline |
ministry |
Procuring ministry/department |
buyer_org |
Buying organization |
bid_type |
BID / RA (Reverse Auction) / Service BID |
status |
open / closed / awarded |
detail_url |
Full URL to bid detail page |
is_relevant |
Auto-tagged if matches drone/UAV keywords |
GeM Portal URLs (Reverse-Engineered)
BASE: https://bidplus.gem.gov.in
Listings:
/all-bids → All bids (paginated)
/bidlists → Ongoing bids
/bidresultlists → Completed bids
/advance-search → Search (POST)
Detail:
/bidding/bid/getBidResultView/{id}
Search POST params:
search_bid, ministry, org_name, bid_no, date_from, date_to
Rate Limiting
- Default: 1.5s between requests (configurable via
REQUEST_DELAY_SECenv var) - GeM is a public portal — public bid data is freely accessible
- Don't run concurrent crawls to avoid IP blocking
Files
gem_mcp/
├── server.py ← MCP server (main entry point)
├── gem_scraper.py ← Scraping logic + MongoDB operations
├── requirements.txt ← Python dependencies
├── Dockerfile ← Container for Render
├── render.yaml ← Render blueprint
├── .env.example ← Environment variable template
├── .gitignore ← Git ignore rules
└── README.md ← This file
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。