Documentation

Everything you need to build, extend, and run an OpenRoutines agent.

Start here

  • Getting started – A quick start guide for building, running, and deploying your first agent.
  • Core concepts – How the pieces of an agent fit together: routines, what a run can touch, the knowledge they share, teamwork, and the supervisor that runs them.

Building

  • Local development – Working on routines on your own machine: what Docker gives you, the write-rehearse-run loop, fixtures, and how a change reaches production.
  • Creating routines – The frontmatter reference for routines, plus scheduling, triggers, MCP servers, best practices, and how a routine's work gets recorded.
  • Extending your agent – Adding capability to an agent: skills, credentials and typed credentials, variables, models and custom endpoints, and plugins.

Running

  • Deploying your agent – Running an agent in production: the container and its two secrets, operational properties, logs, continuous deployment, and framework updates.

Reference

  • CLI – Every OpenRoutines command, what it does, and the flags that change how it behaves.
  • Configuration – Every key in openroutines.yml: what it declares, whether it's required, and what it defaults to.
  • Knowledge – What an agent records as it works: events, tasks, context, per-routine ledgers, and the git branch that holds them.
  • Teamwork – How an agent declares intentions, reports progress, calls out blockers, and tracks tasks, and what each takes in a routine.