Getting startedUse an AI coding assistant

Use an AI coding assistant

AI coding agents such as Claude Code, Cursor and Codex don’t know Memgraph well by default. Give them Memgraph-specific knowledge and access to the current docs, so they write correct Cypher and follow Memgraph’s recommended practices.

Choose what to give your agent

You can combine these, depending on what the agent needs:

Your agent needs toUse
Write Cypher, model data and use MAGE correctly, with knowledge it loads when a task needs itAgent Skills
Look up the current docs and answer with sourcesDocs MCP server
Read the docs without installing anythingllms.txt and Markdown pages
Query your own Memgraph databaseMemgraph MCP server
Remember what it did in earlier sessionsAI memory, which records your agent’s sessions in Memgraph

Machine-readable docs

Agents can read the docs directly, without rendering the website:

  • llms.txt: https://memgraph.com/docs/llms.txt lists the main pages of each docs section with a one-line description, in the same order as the sidebar. Reference material that an agent can skip when its context is short, such as release notes, is grouped under Optional.
  • Section llms.txt files: each section has its own llms.txt with all of its pages, for example https://memgraph.com/docs/querying/llms.txt. The root llms.txt links to them.
  • Markdown pages: add .md to any docs URL to get the page as Markdown, for example https://memgraph.com/docs/querying.md. Every link in the llms.txt files points at the Markdown version.

The rest of the Memgraph website has its own index at https://memgraph.com/llms.txt.

To point your agent at the docs, add the llms.txt URL to its project instructions, for example CLAUDE.md, AGENTS.md or Cursor rules:

Memgraph documentation: https://memgraph.com/docs/llms.txt
First steps with DockerAgent Skills