GDELT MCP
A remote, read-only MCP server for global media intelligence using GDELT DOC 2.0 API, enabling search and analysis of multilingual news coverage, tone, and attention trends.
README
GDELT MCP
A remote, read-only Model Context Protocol server for global media intelligence using the public GDELT DOC 2.0 API.
The server requires no GDELT API key and is deployed on Vercel.
Tools
| Tool | Purpose |
|---|---|
search_articles |
Search multilingual global news coverage and return structured article results |
get_coverage_timeline |
Track normalized media attention or raw matching article counts over time |
get_tone_timeline |
Track the average tone of matching coverage over time |
get_source_country_timeline |
Compare attention across countries where publishing outlets are based |
The documented GDELT GEO API currently returns HTTP 404, so geographic mapping is deliberately not exposed as a tool.
Important interpretation note
GDELT is a large-scale automated media-monitoring system. Its translation and tone analysis can contain errors. Results should be interpreted as signals about news coverage, not as verified facts or direct measurements of real-world events.
Reliability controls
GDELT currently asks API users to keep requests at least five seconds apart. This server therefore:
- serializes outbound GDELT requests at least 5.5 seconds apart per warm instance;
- limits each upstream attempt to 15 seconds;
- retries HTTP 429 once after at least six seconds;
- respects a longer
Retry-Afterheader when supplied; - caches successful identical requests for five minutes; and
- limits response size and MCP request volume.
Because Vercel uses shared serverless egress, GDELT may still occasionally return HTTP 429 during periods of heavy traffic. The MCP returns that error transparently instead of fabricating data.
Architecture
- Plain Node.js with no runtime dependencies
- Vercel serverless functions in the Dubai region
- Stateless MCP Streamable HTTP
- Primary MCP endpoint:
/mcp - Alternate MCP endpoint:
/api/mcp - Health endpoint:
/health - Optional deep DOC check:
/health?deep=1
Local validation
npm install
npm test
npm run build
Local development with Vercel CLI
npm install -g vercel
vercel dev
Then connect an MCP client to:
http://localhost:3000/mcp
Deploy to Vercel
- Import this repository into Vercel.
- No environment variables are required.
- Deploy and open
/health. - Connect the MCP client to
https://<your-domain>/mcp.
Optional environment variables
| Variable | Purpose |
|---|---|
GDELT_DOC_API_BASE |
Override the public DOC endpoint for testing |
MCP_ALLOWED_ORIGINS |
Comma-separated allowed browser origins |
MCP_RATE_LIMIT_PER_MINUTE |
MCP requests per IP per minute; defaults to 30 |
No secrets should be committed.
Example prompts
- “Search global coverage of sovereign AI published in the last 24 hours.”
- “Compare how much attention different publishing countries are giving to data-centre power constraints.”
- “Show the seven-day coverage-volume and tone trends for NVIDIA export restrictions.”
- “Find the latest Arabic-language coverage of AI regulation.”
GDELT query examples
The query field supports GDELT syntax, including:
"artificial intelligence"
(climate OR flooding OR drought)
near20:"Trump Putin"
theme:ECON_INFLATION
tone<-5
Structured tool arguments can also add source-country, language, exact-domain, theme and tone filters.
MCP client configuration
{
"mcpServers": {
"gdelt": {
"url": "https://<your-domain>/mcp"
}
}
}
License
MIT
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。