Replace an agent
Swapping an agent across a run's future work never rewrites what already happened on the agent it's replacing.
AgentDock Pro
Replacing an agent across a run is part of AIgency, available only in AgentDock Pro.
If an agent isn't working out partway through a run, you can replace it for the rest of the run's eligible work — without falsifying what already happened.
What moves, and what doesn't
- Completed, skipped, or stopped tasks stay attributed to whichever agent actually ran them — replacement never rewrites history.
- A task currently running on the old agent is allowed to finish on that agent; it isn't killed just because the run-level replacement changed.
- Pending tasks, and tasks in a failure or capacity-wait state, can move to the new agent.
- Fallbacks, default assignments, permission overrides, and reviewer criteria cascade to reflect the replacement where they referenced the old agent.
Why this matters
This is the same historical-integrity principle behind run history generally: a completion report or timeline entry should always reflect which agent genuinely did the work, even after you've moved the rest of the run to someone else.