Structured Agent Exchange
Orientation
SAE gives agents a shared way to describe the work before transport and runtime details take over. The aim is less renegotiation, clearer handoff, and outputs that are easier for another system or a human reviewer to understand.
SAE is intentionally small. If a field does not help preserve task meaning across systems, it probably does not belong in the public core.
Concept
SAE focuses on objective, constraints, deliverable, delegation hints, review posture, and instruction framing.
Translation
Different model families benefit from different prompt tactics. SAE keeps the packet stable while the translator adapts presentation for the endpoint.
Proof
Adoption claims should be earned through conformance, trace-based comparisons, and at least one outside implementation trial.
Reading path