Skip to main content
Experimental API - Context Connectors is experimental and subject to breaking changes.
Build indexes from Git repos (GitHub, GitLab, BitBucket) or documentation websites. Index code, documentation, runbooks, schemas, configs, and more. Use DirectContext in the Context Engine SDK for custom sources. Store indexes on your local filesystem for fast & simple access, or in S3 for persistent storage in production apps. Search indexes via interactive agent, MCP for AI integrations, CLI for quick searches, or DirectContext in the Context Engine SDK for custom implementations.

​
Getting Started

New to Context Connectors? Here’s a suggested path:
  1. Index and search → Index and Search Code (3 min)
  2. Connect an agent → Local MCP Server (3 min)
  3. Build an app → Store Indexes in S3 (5 min)
  4. Automate updates → GitHub Actions Auto-Indexing (5 min)
Or jump directly to any recipe below.

​
Quickstart Recipes

Index and Search Code

Index a Git repository and search or chat with it (3 minutes)

Index Websites

Crawl and index documentation sites (3 minutes)

Local MCP Server

Add codebase search to Claude Desktop, Claude Code, Cursor, or another agent (3 minutes)

Remote MCP Server

Expose indexes via MCP over HTTP for remote clients (3 minutes)

Store Indexes in S3

Store indexes in S3 for persistent storage in production apps (5 minutes)

GitHub Actions Auto-Indexing

Automatically re-index on every push using GitHub Actions (5 minutes)

Auto-Index with Webhooks

Custom webhook server for auto-indexing (advanced, 10 minutes)

​
Advanced Recipes

Custom Indexer

Build a custom indexer for any data source using DirectContext

Custom Store

Create custom storage backends (local, S3, or any storage)

Custom Client

Build custom search clients for your applications

​
Documentation

CLI Reference

Complete command-line reference