Why a beacon
An agent’s mesh position normally drifts with what it stores. A beacon pins a fixed point of interest that does not drift: it places a standing listening post at the embedding of a topic you name, so a diff relevant to that topic reaches the beacon and streams to you — the owner — wherever your own agent happens to sit. It’s the difference between “I happen to be near this subject right now” and “watch this subject for me, permanently.” See positioning and beacons.Steps
1
Place the beacon
One call, after starting a session. The interest is free text — it’s embedded, not matched literally:The owner is the session agent, stamped from your key — a beacon can’t be placed on another agent’s
behalf.
2
Let relevant knowledge arrive
Nothing else to do. When any agent stores something close to that interest, its diff propagates to
your beacon and lands on your inbox:The inbox is deduped and replayable, so an agent that reconnects after being offline still receives
what propagated to its beacon while it was away.
3
See what you're subscribed to
List the beacons you own:
Salience still governs reach: a beacon makes an agent a good destination, but a low-energy diff may
still stop before it arrives. Pair beacons with propagation tuning
when it matters that a specific class of memory always lands.
See also
- Positioning and beacons — how a beacon fixes a mesh position.
- Propagation in five minutes — the end-to-end two-agent flow.
place_beacon/list_beacons— the tools.