A Model Context Protocol (MCP) server for integrating Jina.ai's Grounding API with LLMs. This server provides efficient and comprehensive web content grounding capabilities, optimized for enhancing LLM responses with factual, real-time web content.
- 🌐 Advanced web content grounding through Jina.ai Grounding API
- 🚀 Real-time content verification and fact-checking
- 📚 Comprehensive web content analysis
- 🔄 Clean format optimized for LLMs
- 🎯 Precise content relevance scoring
- 🏗️ Built on the Model Context Protocol
This server requires configuration through your MCP client. Here are examples for different environments:
Add this to your Cline MCP settings:
{
"mcpServers": {
"jinaai-grounding": {
"command": "node",
"args": ["-y", "mcp-jinaai-grounding"],
"env": {
"JINAAI_API_KEY": "your-jinaai-api-key"
}
}
}
}
For WSL environments, add this to your Claude Desktop configuration:
{
"mcpServers": {
"jinaai-grounding": {
"command": "wsl.exe",
"args": [
"bash",
"-c",
"JINAAI_API_KEY=your-jinaai-api-key npx mcp-jinaai-grounding"
]
}
}
}
The server requires the following environment variable:
JINAAI_API_KEY
: Your Jina.ai API key (required)
The server implements MCP tools for grounding LLM responses with web content:
Ground LLM responses with real-time web content using Jina.ai Grounding.
Parameters:
query
(string, required): The text to ground with web contentno_cache
(boolean, optional): Bypass cache for fresh results. Defaults to falseformat
(string, optional): Response format ("json" or "text"). Defaults to "text"token_budget
(number, optional): Maximum number of tokens for this requestbrowser_locale
(string, optional): Browser locale for rendering contentstream
(boolean, optional): Enable stream mode for large pages. Defaults to falsegather_links
(boolean, optional): Gather all links at the end of response. Defaults to falsegather_images
(boolean, optional): Gather all images at the end of response. Defaults to falseimage_caption
(boolean, optional): Caption images in the content. Defaults to falseenable_iframe
(boolean, optional): Extract content from iframes. Defaults to falseenable_shadow_dom
(boolean, optional): Extract content from shadow DOM. Defaults to falseresolve_redirects
(boolean, optional): Follow redirect chains to final URL. Defaults to true
- Clone the repository
- Install dependencies:
pnpm install
- Build the project:
pnpm run build
- Run in development mode:
pnpm run dev
- Update version in package.json
- Build the project:
pnpm run build
- Publish to npm:
pnpm run release
Contributions are welcome! Please feel free to submit a Pull Request.
MIT License - see the LICENSE file for details.
- Built on the Model Context Protocol
- Powered by Jina.ai Grounding API