implicit-coordination

Stigmergy enables agent coordination by encoding signals in a shared environment rather than passing messages. Agents write to and read from a DuckDB-backed trace store, using seeds and trits to orchestrate actions across generator, coordinator, and validator roles with a single causality trip.

Stigmergy is indirect coordination through environment modification. Instead of message exchanges, agents leave traces that other agents read. This approach eliminates multi-hop messaging delays, achieving low-latency coordination in one causality round while maintaining causal history through seed-chain semantics.

Yes. DuckDB serves as the backing store for environment traces in stigmergic coordination. Agents write seed and trit data to DuckDB, and other agents query the trace to read coordination signals, enabling role-based task distribution with GF(3) conservation checks and triad interactions.

Choose environment-based coordination when agents operate in distributed setups where latency matters and coordination can be expressed through state modifications. It fits seed generation, verification, and role assignment workflows where agents benefit from reading a shared trace instead of managing message queues.

Stigmergy requires all agents to access a shared environment store and understand the encoding scheme (seeds, trits, colors, and causal semantics). It works best for coordinated workflows with clear role separation; complex multi-round negotiations may not fit the single-causality-trip model.

Seeds encode causal history in the trace, and seed-chain semantics enforce ordering and dependencies. Validators read the seed chain to verify generator and coordinator actions satisfy GF(3) conservation constraints, ensuring correctness without explicit message handshakes.