Skip to main content
A runtime cluster runs the releases you make in the Lab, so that other systems can call your blueprints. This guide is for developers and integrators who deploy a release and call it. The guide has these pages, in this order. To try the whole way once, from a deploy to a first call, start with Deploy a release and call it.
  1. What is Leapter: what Leapter does, and the words this guide uses.
  2. Runtime clusters, environments and releases: the parts of a runtime cluster.
  3. Deploy a release and call it: a first deploy from the Lab, a key, and a first call.
  4. Create and revoke an evaluation key: give a calling system access, and take it back.
  5. Call a release over REST: run a blueprint with an HTTP request.
  6. Connect an MCP client to a release: give an AI agent your blueprints as tools, in Claude Code, Claude Desktop, Cursor or another client.
  7. Call a release from n8n with the Leapter node: install the Leapter node in a self-hosted n8n and run a blueprint with it.
  8. Call a release from n8n: import a ready-made n8n workflow that calls the runtime cluster.
  9. Run a blueprint inside n8n: the Code workflow, which runs the logic in n8n without calling the runtime cluster.
  10. Use your blueprints in an n8n AI agent: the MCP workflow, an agent with your blueprints as tools.
  11. Find a run and read usage: look up one call and see how much ran.