ProtocolExperimentalA2A

a2a-agent-registry-on-aws

Name: a2a-agent-registry-on-aws
Rating: 3.0 (28 reviews)
Author: awslabs

by awslabs

A2A AgentCard registry with semantic search on AWS serverless

Python

Updated Oct 19, 2026

Stars

Forks

Commits/Week

Commits/Month

View on GitHub

Overview

Implements a scalable agent registry that indexes AgentCard-compatible A2A agent metadata and supports semantic search. Runs as an AWS serverless stack (Lambda, S3 vector store, Bedrock) with a Python SDK and a React web UI for discovery. Distinctive features include AgentCard A2A compliance and a vector-backed lookup for agent track records and capabilities, leveraging a Vector database for fast retrieval and referencing the Model Context Protocol (MCP) Pattern.

The Value Proposition

As agents begin to discover and call other agents, having a reliable registry of who does what and how they've performed is essential for trust and routing. This project provides an infrastructure piece that surfaces agent track records and searchable capability metadata, enabling evaluation pipelines and trust signals to attach to real agent identities. Until registries like this are common, it’s hard to build reputation-aware delegation and continuous A2A evaluation workflows. See how it aligns with the Capability Attestation Pattern.

Ideal For

Teams building A2A-capable multi-agent systems who need a searchable registry of agent metadata and track records on AWS. This fits well with orchestrated workflows described in the Orchestrator-Worker Pattern for scalable agent coordination.

Applications

Indexing and discovering A2A-compatible agents by capability and provenance
Providing searchable agent metadata and track records for reputation-aware routing
Backing evaluation pipelines that attach performance history and trust signals to agents

Test in the Playground

Validate against scenarios