Fetch.ai has shipped a tool that could quietly reshape how autonomous AI agents talk to each other across platforms. The A2A Outbound Adapter, part of the project’s uAgents-adapter package, gives developers a straightforward way to connect Fetch.ai’s native agents with any external agent that speaks the A2A (Agent-to-Agent) Protocol.
The A2A Protocol is an interoperability standard notably linked to Google Cloud systems. It defines how independent AI agents find each other, exchange messages, and coordinate tasks. Fetch.ai’s new adapter bridges the gap between its own agent ecosystem, which includes both uAgents and Agentverse agents, and the broader universe of A2A-compatible systems.
Two configuration modes are available. The SingleA2AAdapter handles connections to one external A2A server, while the MultiA2AAdapter supports simultaneous connections to multiple servers with multi-port operation. Under the hood, the adapter includes chat protocol forwarding, automatic agent discovery and registration, health monitoring, and routing strategies like keyword matching.
The adapter requires Python 3.10 or higher and is available in versions 0.6.0 and 0.6.2 of the uAgents-adapter package. Documentation lives on both innovationlab.fetch.ai and uagents.fetch.ai.
