Agent field notes
How an AI agent handoff works
Follow a request from project and harness selection to a broker flight, completion evidence, and a follow-up with the same handle.
- Formats
- Landscape
- Runtime
- 52 seconds at 30 fps
- Delivery
- MP4 · H.264 · AAC · BT.709
- Source
- openscout.app
Transcript
An agent handoff needs more than a prompt. It needs a target, a task, and a result you can verify.
In OpenScout, use ask when you expect work or a reply. Choose the project and harness explicitly.
The broker resolves a compatible worker and records the request. It tracks that work as a flight.
A receipt shows the request was recorded. Completion needs its own evidence. Keep those two signals separate.
If the caller stops waiting, the work can still be running. Inspect the existing flight before launching the same task again.
Keep the returned handle for inspection and follow-up. That preserves continuity, instead of guessing a new agent name.
Choose a target, track the work, and inspect the outcome. OpenScout is built for high-trust local developer pilots.
Educational diagrams with synthesized English narration. Captions are included in the picture; the optional caption track repeats the same text.
Try the explanation
Sources
Reviewed 2026-10-02. The diagrams illustrate the documented model; they are not a recording of a live agent run.