Connect one agent
Create a workspace and register an agent. Keep the agent-scoped credential in your runner’s secure configuration. Start with the Python or Node.js SDK, or use the documented HTTP contract.
THE SUPERVISION LOOP
Your runtime. A clearer operating routine.
Connect an existing agent, report its work, review the outcome, and add explicit control points where they matter.
01Your agents
Local · cloud · scheduled
02Toolcaise
Activity · incidents · outcomes
03Your decision
Review · approve · investigate
04Your runner
Handle the decision at a checkpoint
Make its work visible
Confirm the first heartbeat, then report a complete run with meaningful steps and an outcome. Report the expected schedule if the task recurs. A heartbeat alone does not establish that the work happened.
Review the result
Open the reported artifact, compare it with the brief, and record a human acceptance or rejection. Investigate incidents using the run history and the evidence in the source system.
Add a decision before action
Instrument compatible SDK checkpoints and guarded operations. Test waiting, rejection, approval, and acknowledgement with harmless actions before relying on the integration.
Evaluate the pilot
Choose your own measures: reported run coverage, reviewed outcomes, and time to investigate incidents. Agree rollout scope and pricing with the team before a paid expansion.
Outbound reporting. Deliberate controls.
Toolcaise receives the metadata your agents send. It does not automatically discover every agent, host arbitrary customer code, or interrupt an external action already in flight.
Open the technical connection guide →START WITH ONE WORKFLOW
Put your next run
under supervision.
Connect the agent you already run. Build confidence from the work you can actually review.