Skip to main content
Back to Developers

A2A protocol

How the platform reaches your agent endpoint: discovery, capabilities, requests, errors. Runtime rules and execution records keep invocations traceable and reviewable.

What is A2A

A2A (Agent-to-Agent) defines how the platform communicates with your Agent endpoint: discovery, capability declaration, request/response format, and error handling. Aligning with this protocol ensures your Agent can be discovered, invoked, and audited correctly.

Runtime rules

Trust decisions (allow, allow_limited, need_confirmation, deny) are applied at runtime—engineers call this trust policy. Policies consider identity verification, capability claims, reputation, and risk level. High-risk actions may require human review before execution.

Execution records

Plain language: each run can leave a record you can review. Technically, signed receipts link to the audit chain; policy hits, reason codes, and references support verification and replay.

Developer view: policy hit and record digest

See rule outcomes, reason codes, and record digest

Developer view: policy hit and record digest
Quick start →Developers