A2A MCP Server
Enables Claude to connect and interact with A2A Protocol agents, allowing discovery of agent capabil...
Open the FastMCP connection interface
- Click the "Install Now" button to open the FastMCP / connection UI where environment variables (ENVs) are entered.
Add the required environment variable: A2A_AGENT_CARDS
- In the FastMCP ENVs panel add a variable named
A2A_AGENT_CARDS. - The value must be a JSON-stringified list of objects; each object must include a
urlkey (the full Agent Card URL) and may include an optionalcustom_headersobject for authentication headers. Example value (paste this whole string as the ENV value):[{"url":"https://example.com/.well-known/agent-card.json","custom_headers":{"X-API-Key":"YOUR_API_KEY"}}] - You can include multiple agents by adding multiple objects in the list (comma-separated). (Format is described in the project README.)
- In the FastMCP ENVs panel add a variable named
How to obtain an Agent Card URL
- Option A — Use a public agent: copy the Agent Card URL provided by the agent registry or provider (for example a2anet listings or a demo repository).
- Option B — If you run the agent yourself: the agent must publish its Agent Card JSON at a well-known path such as
https://<your-domain>/.well-known/agent-card.json. Ensure your agent service is reachable at that URL. (a2a-protocol.org)
If the agent requires authentication (API key / token)
- Get the API key or token from the agent provider (their dashboard or account settings).
- Place it inside
custom_headersin theA2A_AGENT_CARDSlist, e.g.[{"url":"https://example.com/.well-known/agent-card.json","custom_headers":{"Authorization":"Bearer YOUR_TOKEN"}}] - Use the exact header name the agent expects (e.g.,
X-API-Key,Authorization).
Verify the Agent Card URL before saving
- From your terminal, run:
- curl:
curl -sS https://example.com/.well-known/agent-card.json - (optional) format with jq:
curl -sS https://example.com/.well-known/agent-card.json | jq .
- curl:
- The returned JSON should contain top-level metadata such as
name,description,url, and askillsarray (this confirms the card is valid). (a2a-protocol.org)
- From your terminal, run:
Paste the JSON value into FastMCP and save
- In the FastMCP ENVs UI (opened from "Install Now"), paste the JSON-stringified list as the value for
A2A_AGENT_CARDS. - Save / Apply the settings in the FastMCP interface.
- In the FastMCP ENVs UI (opened from "Install Now"), paste the JSON-stringified list as the value for
Start or restart the MCP server
- Use the FastMCP UI’s install/start action (or restart the MCP) so the server picks up the new ENV.
- After startup, use the MCP tools
list_available_agentsto confirm the agent(s) are discoverable.
Troubleshooting tips
- If
list_available_agentsfails to show the agent, re-check:- The Agent Card URL is reachable from the host (curl from the host).
- Any
custom_headersare correct and accepted by the agent endpoint. - The Agent Card JSON is valid and contains
skillsandurl. (a2a-protocol.org)
- If
(Optional) Multiple agents
- Provide a list with multiple agent objects in the same
A2A_AGENT_CARDSvalue:[{"url":"https://agent1.example/.well-known/agent-card.json"},{"url":"https://agent2.example/.well-known/agent-card.json","custom_headers":{"X-API-Key":"KEY2"}}] - Save and restart as above.
- Provide a list with multiple agent objects in the same
Done
- Once the env is saved and the MCP restarted, use
list_available_agentsandsend_message_to_agentto interact with the configured agents.
- Once the env is saved and the MCP restarted, use
Quick Start
Choose Connection Type for
Authentication Required
Please sign in to use FastMCP hosted connections
Run MCP servers without
local setup or downtime
Access to 1,000+ ready-to-use MCP servers
Skip installation, maintenance, and trial-and-error.
No local setup or infra
Run MCP servers without Docker, ports, or tunnels.
Always online
Your MCP keeps working even when your laptop is off.
One secure URL
Use the same MCP from any agent, anywhere.
Secure by default
Encrypted connections. Secrets never stored locally.
Configuration for
Environment Variables
Please provide values for the following environment variables:
HTTP Headers
Please provide values for the following HTTP headers:
started!
The MCP server should open in . If it doesn't open automatically, please check that you have the application installed.
Copy and run this command in your terminal:
Make sure Gemini CLI is installed:
Visit Gemini CLI documentation for installation instructions.
Make sure Claude Code is installed:
Visit Claude Code documentation for installation instructions.
Installation Steps:
Configuration
Installation Failed
More for Communication
View All →
Slack
Experience the most powerful MCP server for Slack workspaces with advanced message and channel support. This feature-rich server allows fetching messages from channels, threads, DMs, and group DMs with embedded user info and smart history pagination by date or count. It supports enterprise workspaces, caching for faster access, and flexible transports like Stdio and SSE with proxy settings. Designed for stealth operation or secure access via OAuth, it integrates channel and thread lookups by name or ID to give rich, contextual Slack workspace data efficiently and reliably. Perfect for developers needing robust Slack data handling.
More for Developer Tools
View All →
GitHub
Extend your developer tools with the GitHub MCP Server—a powerful Model Context Protocol server enhancing automation and AI interactions with GitHub APIs. It supports diverse functionalities like managing workflows, issues, pull requests, repositories, and security alerts. Customize available toolsets to fit your needs, enable dynamic tool discovery to streamline tool usage, and run the server locally or remotely. With read-only mode and support for GitHub Enterprise, this server integrates deeply into your GitHub ecosystem, empowering data extraction and intelligent operations for developers and AI applications. Licensed under MIT, it fosters flexible and advanced GitHub automation.
Desktop Commander
Desktop Commander MCP transforms Claude Desktop into a powerful AI assistant for managing files, running terminal commands, and editing code with precision across your entire system. It supports in-memory code execution, interactive process control, advanced search and replace, plus comprehensive filesystem operations including reading from URLs and negative offset file reads. With detailed audit and fuzzy search logging, it enables efficient automation, data analysis, and multi-project workflows—all without extra API costs. Designed for developers seeking smarter automation, it enhances productivity by integrating all essential development tools into a single, intelligent chat interface.
Chrome DevTools
Provides direct Chrome browser control through DevTools for web automation, debugging, and performance analysis using accessibility tree snapshots for reliable element targeting, automatic page event handling, and integrated performance tracing with actionable insights.
FreeCAD
Enables AI-driven CAD modeling by providing a remote procedure call (RPC) server that allows programmatic control of FreeCAD, supporting operations like creating documents, inserting parts, editing objects, and executing Python code for generative design workflows.
Similar MCP Servers
ROS MCP Server
Unlock true AI-powered robotics—control and observe any ROS1 or ROS2 robot using natural language. ROS-MCP-Server bridges large language models like Claude, GPT, and Gemini with your robot, enabling seamless two-way communication with zero changes to the robot’s code. Instantly browse topics, read sensors, call services, and adjust parameters—all through intuitive commands. Perfect for debugging, real-time control, and monitoring, this server works with both simulated and real robots and supports all MCP-compatible language models. Future updates will add enhanced permissions and action support for even greater flexibility.
Circuitry MCP Server
Connects AI coding agents to Circuitry's visual workflow platform, enabling them to create and sync code nodes from project files, understand user-drawn flowcharts and diagrams, generate visual flowcharts, and create data visualizations like spreadsheets and charts.
DollhouseMCP
A comprehensive Model Context Protocol (MCP) server that enables dynamic AI persona management with an integrated GitHub-powered marketplace, allowing Claude and other compatible AI assistants to activate different behavioral personas.
astro-airflow-mcp
An MCP server that enables AI assistants to interact with Apache Airflow's REST API for DAG management, task monitoring, and system diagnostics. It provides comprehensive tools for triggering workflows, retrieving logs, and inspecting system health across Airflow 2.x and 3.x versions.
Intercom
Intercom hosts a remote MCP server that follows the authenticated remote MCP specification. This server handles requests from AI tools and provides access to Intercom data through a secure interface. When an AI tool or application needs to access Intercom data: The tool connects to Intercom's MCP server; Authentication verifies the user's permissions; The tool can then access relevant Intercom data and functionality; The connection remains live to receive updates as needed.