Use XMTP documentation with AI coding assistants

To make it easier for you to build with XMTP using LLMs and coding assistants, we offer three AI-ready solutions. Choose the ones that best fit your workflow:

Use Claude Code skills if:

You're using Claude Code to build XMTP agents
You want to scaffold, deploy, and troubleshoot agent projects with slash commands
You prefer hands-on automation over reading docs

Use the documentation MCP server if:

You're using MCP-compatible tools like Claude Code, Cursor, or Copilot
You want dynamic, structured documentation querying while writing and debugging code
You're comfortable setting up and running a local server process

Use llms.txt files if:

You're using AI tools that don't support MCP
You want zero-setup access to documentation
You prefer a simple copy-paste or URL reference approach

Claude Code skills

If you use Claude Code, you can add XMTP-specific skills to your agent project. Skills are instruction files that teach Claude Code how to perform tasks like scaffolding a new agent, deploying to Railway, or diagnosing common errors.

Install the skills

Copy the skills into your project's .claude/skills/ directory:

# From your agent project root
mkdir -p .claude/skills
curl -sL https://docs.xmtp.org/skills/create-xmtp-agent.md -o .claude/skills/create-xmtp-agent.md
curl -sL https://docs.xmtp.org/skills/deploy-xmtp-agent.md -o .claude/skills/deploy-xmtp-agent.md
curl -sL https://docs.xmtp.org/skills/xmtp-agent-troubleshoot.md -o .claude/skills/xmtp-agent-troubleshoot.md

Once installed, Claude Code auto-discovers them in your project.

Available skills

Skill	What it does
`create-xmtp-agent`	Scaffolds a new agent project with your choice of brain (Claude, OpenAI, or rules-based) and the XMTP Agent SDK for messaging
`deploy-xmtp-agent`	Deploys your agent to Railway with the correct Dockerfile, environment variables, and persistent storage
`xmtp-agent-troubleshoot`	Diagnoses common errors like `transport error`, wallet key format issues, and installation limit warnings

These skills follow the same three-layer architecture described in the tutorial: your brain, the XMTP messaging framework, and the glue that connects them.

Documentation MCP server

You can run a local XMTP documentation MCP server that you can use with Claude Code, Cursor, Copilot, and other MCP-compatible AI coding assistants.

To learn about how it's built and how to use it, see the xmtp-docs-mcp repo.

llms.txt files

If you're using an AI coding assistant that allows custom context, you can upload or point to the appropriate file to enhance your development experience with XMTP. Using the file focused on your builder use case (chat apps or agents) typically provides better AI performance by narrowing the context.

For building chat apps

Use llms-chat-apps.txt: https://docs.xmtp.org/llms/llms-chat-apps.txt

Includes: Core messaging, content types, push notifications, user consent, protocol fundamentals, network operations, and funding information.

For building agents

Use llms-agents.txt: https://docs.xmtp.org/llms/llms-agents.txt

Includes: Agent concepts, building and deploying agents, agent-specific content types, protocol fundamentals, network operations, and funding information.

For comprehensive coverage

Use llms-full.txt: https://docs.xmtp.org/llms/llms-full.txt

Includes: All documentation for chat apps, agents, protocol fundamentals, network operations, and funding information.