Connect Your Data with Context7

Context7 specializes in connecting code repositories and technical documentation to your agents. It's particularly well-suited for developers who want their agents to have access to library documentation, API specs, and code repositories.

Supported data sources

With Context7 you can connect:

• Code Repositories: GitHub Repositories, GitLab, BitBucket
• API Documentation: OpenAPI Spec
• Documentation Formats: LLMs.txt
• Websites: Website crawling and indexing

Getting started

Step 1: Check if your library is already available

Context7 maintains a library of pre-indexed documentation for popular libraries and frameworks. Before creating an account, check if your library is already available:

Visit context7.com and browse the list of available libraries. If your library is listed, you can use it immediately without additional setup.

Step 2: Create an account

If your library isn't listed or you want to connect custom sources:

1. Sign up for Context7
2. Complete the account setup process
3. Verify your email address if required

Step 3: Connect your data sources

1. Log in to your Context7 dashboard
2. Add your repositories, websites, or OpenAPI specs
3. Wait for Context7 to index your content
4. Verify that your content is accessible

Step 4: Get the Context7 Library ID

To use a specific library with your agents, you'll need the Context7 Library ID. You can find this ID in the library's URL on context7.com.

How to find the Library ID:

1. Navigate to the library page on context7.com (e.g., https://context7.com/supabase/supabase)
2. Copy the path after the domain name
3. The Library ID is the path portion of the URL

Example:

• URL: https://context7.com/supabase/supabase
• Library ID: supabase/supabase

Step 5: Register the MCP server

Register the Context7 MCP server as a tool in your agent configuration:

Using TypeScript SDK:

import { mcpTool, subAgent } from "@inkeep/agents-sdk";

const context7Tool = mcpTool({
  id: "context7-docs",
  name: "context7_search",
  description: "Search code documentation and library references",
  serverUrl: "https://mcp.context7.com/mcp",
});

const devAgent = subAgent({
  id: "dev-agent",
  name: "Developer Assistant",
  description: "Helps with code questions using library documentation",
  prompt: `You are a developer assistant with access to code documentation.`,
  canUse: () => [context7Tool],
});

Using Visual Builder:

1. Go to the MCP Servers tab in the Visual Builder
2. Click "New MCP server"
3. Enter:
   • Name: Context7 Documentation
   • URL: https://mcp.context7.com/mcp
   • Transport Type: Streamable HTTP or SSE
4. Save the server
5. Add it to your agent by dragging it onto your agent canvas

Step 6: Use the Context7 MCP server in your agent

Once your Context7 MCP server is registered, you can use it with a specific library ID (from Step 4) or let it automatically match libraries based on your query.

Specifying a library ID

If you know which library you want to use, specify its Context7 Library ID in your agent's prompt. This allows the Context7 MCP server to skip the library-matching step and directly retrieve documentation.

Example:

const devAgent = subAgent({
  id: "react-agent",
  name: "React Assistant",
  description: "Helps with React development",
  prompt: `You are a React development assistant. Use library ID "websites/inkeep" 
    when searching for documentation. Use the get_library_docs tool to find React 
    documentation and examples.`,
  canUse: () => [context7Tool],
});

Automatic library matching

If you don't specify a library ID, the Context7 MCP server will automatically match libraries based on your query. The server uses the resolve-library-id tool to identify the appropriate library, then uses the get_library_docs tool to retrieve the relevant documentation.

This approach works well when:

• You're working with multiple libraries
• The library name is mentioned in the user's query
• You want the agent to dynamically select the most relevant library