About AgentRPC
AgentRPC is a universal RPC layer that lets AI agents call functions written in different languages and frameworks. It targets services running in private VPCs, Kubernetes clusters, or multiple clouds, and exposes them through a hosted MCP server and OpenAI-compatible tool definitions.
Key features
- 01SDKs for TypeScript, Go, and Python
- 02Private network registration without open ports
- 03Long polling for calls beyond HTTP timeouts
- 04Tracing, metrics, and event observability
- 05Health tracking with failover and retries
- 06Compatible with MCP and OpenAI SDK agents
Use cases
- 01Letting agents call internal services in private VPCs
- 02Exposing functions across several cloud environments to AI models
- 03Providing remote access to long-running functions
MCP support
AgentRPC offers a hosted MCP server, and its TypeScript SDK can start a local MCP server with npx agentrpc mcp to expose registered tools to MCP clients.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEAgentRPC
Universal RPC layer for AI agents across network boundaries and languages
Overview
AgentRPC allows you to connect to any function, in any language, across network boundaries. It's ideal when you have services deployed in:
- Private VPCs
- Kubernetes clusters
- Multiple cloud environments
AgentRPC wraps your functions in a universal RPC interface, connecting them to a hosted RPC server accessible through open standards:
- Model Context Protocol (MCP)
- OpenAI-compatible tool definitions (OpenAI, Anthropic, LiteLLM, OpenRouter, etc.)
How It Works
- Registration: Use our SDK to register functions and APIs in any language
- Management: The AgentRPC platform (api.agentrpc.com) registers the function and monitors its health
- Access: Receive OpenAPI SDK compatible tool definitions and a hosted MCP server for connecting to compatible agents
Key Features
| Feature | Description |
|---|---|
| Multi-language Support | Connect to tools in TypeScript, Go, Python and .NET (coming soon) |
| Private Network Support | Register functions in private VPCs with no open ports required |
| Long-running Functions | Long polling SDKs allow function calls beyond HTTP timeout limits |
| Full Observability | Comprehensive tracing, metrics, and events for complete visibility |
| Automatic Failover | Intelligent health tracking with automatic failover and retries |
| Framework Compatibility | Out-of-the-box support for MCP and OpenAI SDK compatible agents |
Getting Started
Quick Start
Follow the quick start example on our docs site.
Examples
Explore working examples in the examples directory.
MCP Server
The AgentRPC TypeScript SDK includes an optional MCP (Model Context Protocol) server.
ANGENTRPC_API_SECRET=YOUR_API_SECRET npx agentrpc mcp
This launches an MCP-compliant server for external AI models to interact with your registered tools.
Claude Desktop Integration
Add to your claude_desktop_config.json:
{
"mcpServers": {
"agentrpc": {
"command": "npx",
"args": [
"-y",
"agentrpc",
"mcp"
],
"env": {
"AGENTRPC_API_SECRET": "<YOUR_API_SECRET>"
}
}
}
}
Cursor Integration
Add to your ~/.cursor/mcp.json:
{
"mcpServers": {
"agentrpc": {
"command": "npx",
"args": ["-y", "agentrpc", "mcp"],
"env": {
"AGENTRPC_API_SECRET": "<YOUR_API_SECRET>"
}
}
}
}
License
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
This repository contains all the open-source components and SDKs for AgentRPC.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.