Auto-Generate Agent Cards
Reads your MCP server's tool definitions and automatically generates a standards-compliant A2A Agent Card (JSON-LD). No manual schema writing. Your capabilities are discoverable by the entire A2A ecosystem instantly.
MCP Server · npm package
Bridge MCP to the Google A2A Agent Ecosystem
The TRUSS MCP-A2A Gateway wraps any MCP server and exposes it as an A2A-compatible agent. Auto-generates Agent Cards, translates A2A tasks to MCP tool calls, and makes your tools discoverable by LangGraph, CrewAI, Google ADK, and AutoGen.
Individual from $49/mo. Team from $99/mo.
The Problem
The agent ecosystem is fragmented by protocol
MCP is the protocol for Claude and Cursor. A2A is Google's protocol for cross-framework agent communication. Without a bridge, your MCP tools are invisible to LangGraph, CrewAI, Google ADK, and AutoGen — and vice versa. Building in one ecosystem means missing the other.
4+
Major agent frameworks (LangGraph, CrewAI, AutoGen, Google ADK) using the A2A protocol
0
Native bridges between MCP and A2A before TRUSS — manual protocol translation required
10×
Potential audience reach when your MCP tools are callable from the entire A2A ecosystem
The Solution
Auto-bridge between MCP and A2A
TRUSS MCP-A2A Gateway sits between your MCP servers and the A2A ecosystem. Point it at any MCP server config and it automatically generates the A2A bridge — no manual protocol work required.
A2A Task → MCP Tool Call
Receives A2A task requests from any framework, maps them to the correct MCP tool call with the right parameters, executes them against your MCP server, and returns results in A2A format. Bidirectional translation is automatic.
Works with All A2A Frameworks
Any framework implementing the Google A2A protocol can call your tools: LangGraph, CrewAI, Google ADK, AutoGen, and custom agents. One gateway, all frameworks.
Route to Multiple MCP Servers
Configure multiple MCP servers behind a single A2A gateway. Agents discover all available tools in one Agent Card and the gateway routes each task to the correct upstream MCP server.
Authentication Handling
Forwards API keys and auth tokens from A2A task requests to the appropriate MCP server environment. Supports Bearer tokens, API keys, and custom auth headers.
Task Tracking & Logging
Every A2A task request and MCP tool response is logged with correlation IDs. Trace exactly which framework called which tool, with what parameters, and what was returned.
How It Works
From install to cross-framework agent in 60 seconds
-
01
Point at your MCP servers
Install via npx. Provide your existing MCP server config (the same JSON you use for Claude Code or Cursor). The gateway reads tool definitions automatically.
-
02
Gateway auto-generates your Agent Card
The gateway starts an HTTP server and serves your auto-generated Agent Card at
/.well-known/agent.json. Any A2A-compatible framework can discover your tools immediately. -
03
A2A agents call your MCP tools
LangGraph, CrewAI, Google ADK, or any A2A agent sends task requests to your gateway URL. The gateway translates them to MCP tool calls, executes them, and returns results in A2A format. Zero code changes to your MCP servers.
# Start the A2A gateway wrapping your MCP servers
npx truss-mcp-a2a-gateway \
--mcp-config ~/.claude/mcp.json \
--port 8080 \
--api-key $TRUSS_API_KEY
# Agent Card is auto-served at:
# http://localhost:8080/.well-known/agent.json
# LangGraph / CrewAI can now discover and call your tools:
# POST http://localhost:8080/a2a/tasks/send
{
"id": "task-123",
"message": {
"parts": [{ "text": "analyze the slow queries in production" }]
}
}
Pricing
Plans for individuals and teams
Choose the plan that fits your setup.
Individual
$49/mo
For individual developers bridging MCP to the A2A ecosystem.
- Up to 3 MCP servers
- Auto Agent Card generation
- A2A task translation
- All A2A frameworks
- Auth pass-through
- Task logging
Team
$99/mo
For teams with multiple members and environments.
- Unlimited MCP servers
- Auto Agent Card generation
- A2A task translation
- All A2A frameworks
- Multi-environment routing
- Priority support
- Team access management
- Advanced task analytics
FAQ
Frequently asked questions
What is the Google A2A protocol?
Google's Agent-to-Agent (A2A) protocol is an open standard for AI agents to discover and communicate with each other. It defines Agent Cards (JSON-LD capability declarations) and a task/message exchange format. LangGraph, CrewAI, Google ADK, and AutoGen all support A2A, enabling cross-framework agent communication.
What does the MCP-A2A Gateway actually do?
The gateway wraps any existing MCP server and exposes it as an A2A-compatible agent. It auto-generates an Agent Card from your MCP tool definitions, runs an HTTP server that accepts A2A task requests, and translates each task into MCP tool calls — then returns results in A2A format. Zero changes to your existing MCP servers.
Which agent frameworks can call my MCP tools through the gateway?
Any framework implementing the Google A2A protocol: LangGraph, CrewAI, Google ADK, AutoGen, and any custom agent that can send HTTP requests to an A2A endpoint. The gateway is framework-agnostic.
Do I need to modify my existing MCP servers?
No. The gateway reads your existing MCP server config (the same JSON you use for Claude Code or Cursor), discovers tools automatically, and handles all A2A protocol translation. Your MCP servers run unchanged.
Make your MCP tools callable from anywhere
Connect your MCP servers to LangGraph, CrewAI, Google ADK, and AutoGen with a single npm install. No protocol work required.
No usage limits. Cancel anytime.