DevTk.AI

A2A Agent Card Generator

Generate A2A Agent Card JSON for Google's Agent-to-Agent protocol. Define skills, capabilities, and security schemes.

Provider

Capabilities

I/O Modes

Comma-separated: text, text/plain, application/json, image/*

agent.json Preview

{
  "name": "",
  "url": "",
  "version": "1.0.0",
  "protocolVersion": "0.3.0",
  "capabilities": {
    "streaming": false,
    "pushNotifications": false,
    "stateTransitionHistory": false
  },
  "defaultInputModes": [
    "text"
  ],
  "defaultOutputModes": [
    "text"
  ],
  "supportsAuthenticatedExtendedCard": false
}

Deployment: Serve this file at https://yourdomain.com/.well-known/agent.json for automatic agent discovery.

How to Use This Tool

  1. Start by filling in your agent's basic info: name, endpoint URL, version, and description.
  2. Configure capabilities: enable streaming (SSE), push notifications, or state transition history based on your agent's features.
  3. Set default input/output modes — these define what content types your agent accepts and produces (e.g., text, application/json, image/*).
  4. Add skills to describe what your agent can do. Each skill needs a name, description, tags for discoverability, and example prompts.
  5. Optionally add security schemes (Bearer, API Key, OAuth2, OpenID Connect) to require authentication.
  6. Review the JSON preview, validate your Agent Card, then export it. Deploy the file at /.well-known/agent.json on your server.

What is A2A (Agent-to-Agent Protocol)?

A2A (Agent-to-Agent Protocol) is an open standard by Google (April 2025) that enables independent AI agents to discover, communicate, and collaborate with each other. While MCP connects agents to tools, A2A connects agents to other agents — making it possible to build multi-agent systems across organizational boundaries.

The core discovery mechanism is the Agent Card — a JSON file served at /.well-known/agent.json that describes an agent's identity, skills, capabilities, and authentication requirements. Any A2A-compatible system can discover your agent simply by fetching this URL.

A2A defines a rich task lifecycle: tasks move through states like queued, running, input-required, and completed. This makes it ideal for long-running workflows that need progress tracking, human-in-the-loop approval, or cross-agent coordination.

The protocol launched with support from 50+ partners including Salesforce, SAP, ServiceNow, LangChain, and PayPal. Version 0.3 (July 2025) added gRPC support and signed security cards. A2A is governed by the Linux Foundation.

This generator helps you build valid A2A Agent Cards visually. It supports all fields from the v0.3 specification including skills, security schemes, capabilities, and I/O modes. The generated agent.json can be deployed immediately for agent discovery.

Last updated: March 2026

FAQ

What is A2A (Agent-to-Agent Protocol)?

A2A is an open protocol by Google (April 2025) that enables independent AI agents to discover, communicate, and collaborate. It uses Agent Cards (JSON metadata at /.well-known/agent.json) for discovery, and defines a task lifecycle for stateful multi-agent workflows.

What is an Agent Card?

An Agent Card is a JSON file served at /.well-known/agent.json that describes your agent's identity, skills, capabilities, and authentication requirements. Other agents discover your agent by fetching this URL — similar to how robots.txt works for web crawlers.

How is A2A different from MCP?

MCP connects agents to tools (vertical integration), while A2A connects agents to each other (horizontal collaboration). They're complementary: an agent uses MCP internally to access tools, and A2A externally to communicate with other agents.

Which companies support A2A?

A2A launched with 50+ partners including Google, Salesforce, SAP, ServiceNow, PayPal, LangChain, MongoDB, Atlassian, Box, Cohere, Intuit, and UKG. The protocol is governed by the Linux Foundation as an open standard.

Can I use both MCP and A2A?

Yes! They're designed to be complementary. Use MCP inside your agent to access tools and data sources, and use A2A to communicate with other agents externally. This is the recommended architecture for production multi-agent systems.

Where do I deploy the agent.json file?

Serve your Agent Card at https://yourdomain.com/.well-known/agent.json following RFC 8615 (Well-Known URIs). Any A2A-compatible client can discover your agent by fetching this URL. Make sure it's accessible over HTTPS in production.

Related Tools