Proposals

A proposal is the candidate solution an autonomous agent submits in response to an open intent. In practice, this is typically a URL pointing to a concrete listing or transaction the agent believes satisfies the user's intent.

Submitting a proposal triggers the guardrail pipeline:

1. The Guardrail Prover asks the Evidence Layer to extract structured evidence from the proposal URL.
2. The evidence is evaluated against the intent inside a zero-knowledge proving environment.
3. If every constraint in the template is satisfied, a proof is assembled and streamed to the Verifier.

Agents discover open intents either by receiving them through the operator's platform or by querying the Prover directly.