Glossary

    A2A (Agent2Agent)

    Published · Last updated
    Next scheduled review:

    Agent2Agent (A2A) is an open protocol that lets independent AI agents, built by different vendors on different frameworks, discover each other, exchange messages and hand off tasks without sharing their internal memory or tools.

    Google launched A2A on 9 April 2025 with more than 50 technology partners, built on existing standards such as HTTP, server-sent events and JSON-RPC (Google Developers Blog, 9 April 2025). Each agent publishes an Agent Card, a JSON file at /.well-known/agent-card.json, that describes its skills, endpoints and security requirements. Work is organised as tasks with a lifecycle, so long jobs can report progress. The specification defines JSON-RPC, gRPC and HTTP+JSON bindings (A2A Protocol specification, accessed 9 November 2026).

    Version 1.0.0 shipped on 12 March 2026 and 1.0.1 on 28 May 2026 (GitHub, a2aproject/A2A releases). On 27 August 2026 the project became a Growth Stage project of the Linux Foundation's Agentic AI Foundation, alongside MCP. The project's own short version: MCP connects an agent to its tools, A2A connects agents to each other. Ad tech has started slowly: one of the 53 entries in the IAB Tech Lab's agent registry used A2A on 9 November 2026.

    Example: A brand's planning agent reads a publisher agent's Agent Card and sees a skill for proposing media plans. It sends a task with the brief and budget. The publisher agent uses its own MCP tools to check inventory, then returns a proposal as the task's result.