Guides
Each guide answers one question: how do I get this done? They run from start to finish and stay close to the task. Exact commands, flags, fields, and endpoints live in Reference; the supporting mental model lives in Concepts.
The list below is grouped by the job in hand: getting a first result, running work that waits on something, coordinating more than one agent, driving Holon from code or a browser, working with external content, and fixing a task that stalls.
Every guide follows the same shape: goal and context, prerequisites, steps, and how to confirm success.
Run your first Holon taskStart Holon, run one task end to end, and confirm the result.Run a GitHub task with holon solveTake an issue or pull request from input to a finished task, then check the result.Run a long-lived taskStart work that waits, check progress, survive disconnects, and collect the final brief.Delegate work to another agentHand a scoped task to a child agent, wait for the result, and handle what comes back.Automate Holon over HTTPAuthenticate, submit work, follow status, and read results from code.Connect to a remote Holon runtimeReach a runtime running on another machine and verify the connection.Use the Web GUIDrive agents, work items, and skills from the browser.Configure OIDC authenticationSet up OpenID Connect single sign-on, configure session policies, and audit user prompts.Create an agent from a templatePick a template, create an agent, and confirm it can take a task.Add a skill to an agentFind a skill, install it to the library, enable it for an agent, and confirm it is active.
