# ArchGraph Docs > Code intelligence for AI agents. ArchGraph parses your codebase, builds a knowledge graph, and exposes it to AI agents via the Model Context Protocol (MCP). ArchGraph parses 16 languages with tree-sitter + SCIP, imports the structural graph into Neo4j, and exposes 21 MCP tools so AI agents can search, trace call chains, analyze blast radius, and scan for vulnerabilities. This index links every doc page; see [llms-full.txt](https://docs.archgraph.dev/llms-full.txt) for the full corpus as a single markdown file. ## Introduction - [ArchGraph](https://docs.archgraph.dev/): Code intelligence for AI agents. ArchGraph parses your codebase, builds a knowledge graph, and exposes it to AI agents via the Model Context Protocol (MCP). ## Getting Started - [Quick Start](https://docs.archgraph.dev/quick-start): Get ArchGraph running in 5 minutes. - [Connect an Agent](https://docs.archgraph.dev/connect-agent): Set up ArchGraph with your preferred AI coding assistant. ## MCP Tools - [MCP Tools](https://docs.archgraph.dev/mcp-tools): Reference for the 21 MCP tools ArchGraph exposes — code search, graph analysis, security taint, cross-repo impact, and workspace dependency graphs. - [Example Queries](https://docs.archgraph.dev/example-queries): Common questions to ask your AI agent with ArchGraph connected. ## Reference - [API Reference](https://docs.archgraph.dev/api-reference): The ArchGraph REST API is available at `api.archgraph.dev`. - [Supported Languages](https://docs.archgraph.dev/supported-languages): The sixteen languages ArchGraph parses, and the difference between parsing a language and resolving its calls with a compiler. - [What your install analyses](https://docs.archgraph.dev/capabilities): What a deployed ArchGraph build actually analyses — parsing, call resolution and deep analysis are three separate answers. ## Self-hosting - [Self-Hosting](https://docs.archgraph.dev/self-hosting): This page covers running the **hosted product's stack** on your own server: the API, the databases, and the Cloudflare Worker gateway in front of them. Identity still goes through Clerk and the MCP endpoint is still the - [On-premise](https://docs.archgraph.dev/on-prem): Running ArchGraph entirely on your own hardware, with no identity provider and no outbound dependency. - [Upgrading](https://docs.archgraph.dev/upgrade): Moving a self-hosted ArchGraph from one build to the next, and back again. ## Code hosts - [Connect GitHub](https://docs.archgraph.dev/connect-github): Connecting ArchGraph to GitHub — token permissions, connections.yml, GitHub Enterprise, webhooks, and what each failure means. - [Connect GitLab](https://docs.archgraph.dev/connect-gitlab): Connecting ArchGraph to GitLab — which token scope actually works, self-managed instances, webhooks, and what each failure means. ## More - [FAQ](https://docs.archgraph.dev/faq): **What is ArchGraph?** ArchGraph is a code intelligence service that builds knowledge graphs from your codebase and exposes them to AI agents via MCP. Agents can search, trace call chains, analyze impact, and detect vuln ## Related - [ArchGraph marketing site](https://archgraph.dev): product overview, pricing, and changelog. - [Marketing llms.txt](https://archgraph.dev/llms.txt): marketing-side index for AI crawlers. - [GitHub: Deaxu/ArchGraph](https://github.com/Deaxu/ArchGraph): core extraction engine (tree-sitter + SCIP).