Skip to content

Agent assignment

Each AIgency task carries an assignment describing which agent runs it, at what model and reasoning effort, with optional fallbacks.

Every task in a run has an assignment: which agent runs it, which model and reasoning effort to use, and what to do if that agent fails.

AgentDock Pro

Agent assignment is part of AIgency, available only in AgentDock Pro.

Assignment fields

  • Agent, model, reasoning effort — the specific configuration the task runs with.
  • source — how the assignment was made:
    • manual — you assigned it yourself.
    • planner — AIgency's planner chose it, based on task routing.
    • reassigned — it was moved to a different agent after the original one failed or was unavailable.
    • fallback — a fallback agent from the ordered list below took over.
  • Fallback agents — an ordered list of agents to try if the primary assignment can't complete the task.
  • allowReassignOnFailure — an opt-in flag controlling whether AIgency is allowed to move a failed task to a different agent automatically, rather than leaving it failed for you to handle manually.

How assignment decisions get made

By default, AIgency's planner assigns agents using task routing — a capability-matrix scoring system matched against the task's category. You can override any assignment manually before or during a run.

Related

  • Task routing — how the planner scores agents against a task.
  • Retries — what happens when an assigned agent's attempt fails.