What is the Agent Query Protocol?

A shared language for asking agents bounded questions and receiving structured answers.

Overview

Agent Query Protocol (AQP) is an open standard for systems to ask agents bounded, typed questions and receive structured, policy-controlled answers. It gives applications, tools, agents, and people a common query contract without requiring them to share a model provider, framework, transport, memory system, or internal reasoning.

AQP standardizes the meaning and lifecycle of a query exchange. Bindings define how AQP objects travel over HTTP, MCP, A2A, and future transports.

What AQP enables

A tool can ask whether its result satisfied the user after the outcome is known.
An agent can request clarification, approval, evaluation, status, preference, or correction in one portable shape.
A requester can constrain an answer to text, structured data, a choice, a boolean, or a number.
A runtime can enforce local consent, privacy, and authorization policy regardless of what is requested.
Implementers can add new transports without redefining query semantics.

How AQP fits

MCPconnects AI applications to tools, data, and workflows

A2Aconnects independent agentic applications to each other

AQPstandardizes bounded questions and structured answers across either relationship

Design principles

Transport-neutral core. Canonical objects cannot depend on a single framework or vendor.
Bounded by default. Each request names one subject, one prompt, and one permitted response shape.
Local authority. A requester's desired consent mode never overrides respondent, host, organization, or user policy.
Explicit provenance. Subject, respondent kind, confidence, and delivery state remain machine-readable.
Failure isolation. Query delivery must not delay or fail the interaction that carried it.