Skip to content

Playbooks

A playbook is the reusable process definition behind a workflow.

It answers questions like:

  • what outcome are we trying to reach?
  • what stages or work items should exist?
  • which specialists are involved?
  • what should the orchestrator pay attention to?
  • where can operators approve, redirect, or request changes?

In practice, a playbook combines several layers:

  • the outcome and workflow shape
  • prose instructions that explain how the work should be driven
  • role and specialist expectations
  • launch-time parameters and input packet structure
  • lifecycle defaults for planned or ongoing work
  • advanced control settings when a workflow truly needs them

The important part is that the core governance lives in authored process prose, not in a brittle lattice of hidden rules.

Without a playbook, a workflow launch is mostly “prompt and hope.” With a playbook, the platform has a real process definition to execute against.

That gives you:

  • repeatable workflow launches
  • clearer operator expectations
  • reusable patterns across similar work
  • a better boundary between authored process intent and runtime execution

Agirunner also ships a shared public catalog in agirunner-playbooks.

That repository is the source of truth for curated community playbooks, specialists, and skills that operators can import from the dashboard. The import model is deliberately simple:

  • operators choose one playbook at a time
  • the platform fetches that playbook and its referenced specialists and skills
  • imported artifacts become normal tenant-local copies
  • the platform keeps quiet provenance metadata instead of forcing a new community concept into everyday workflow use

That means the catalog is a starting point, not a lock-in mechanism. A team can import a strong workflow package, then adapt it locally as its own operating process.

The dashboard authoring flow supports both:

  • planned work for a bounded run toward a specific outcome
  • ongoing work for boards or recurring operational processes that stay alive over time

That distinction matters because the operator experience, launch flow, and lifecycle expectations differ.

  • use Work Design -> Playbooks in the dashboard to browse and author them
  • use the platform API when you want to create, update, archive, or launch them programmatically
  • use the runtime only after the platform has already turned the playbook into explicit task contracts