What problem does it solve?
When two daemons (e.g., a home Mac Studio and a k8s pod) both declare hosting for the same agent, the binding becomes contested and dispatch routing stays pinned to the original owner. This Skill gives the orchestrator a deterministic way to resolve that contention by explicitly rebinding the agent to a chosen target daemon.
Core Features & Use Cases
- Contested Binding Resolution: Consumes
agent.binding.contested sync events and calls the /api/im/agent-bindings/:agentImUserId/rebind endpoint to atomically flip ownership.
- In-Flight Task Awareness: Returns the in-flight task count on the previous daemon so the orchestrator can wait for draining before declaring the migration done.
- Audit & Sync Propagation: Writes an audit row with the supplied reason and emits
agent.binding.rebound so other daemons drop stale routing state.
- Use Case: A user says "pin my CEO agent to my laptop instead of the k8s pod" — the orchestrator inspects the workspace bindings, posts a rebind with a reason, and reports how many tasks are still finishing on the old daemon.
Quick Start
Ask the orchestrator to resolve the contested binding for your agent by rebinding it to your preferred daemon, for example: "Pin agent u_agent_ceo to daemon-mac-studio-01 because I'm on my home network."