Agent RPC makes your functions available through standards such as Model Context Protocol and OpenAI Tool Calling. Register tools without exposing new API endpoints including within private VPCs, IOT devices and local machines.
Hey Product Hunt! 👋
AgentRPC is a managed platform and set of developer SDKs for connecting agents to functions (tools) across network and language boundaries.
This allows you to wrap functions and make them available to AI agents using Model Context Protocol (MCP) and OpenAI tool calls (including the new Agents SDK) and makes for easier decoupling of tools and agent implementation.
Cases for this include registering tools from services with no inbound network access (Local Machines, IoT, Private VPCs, etc).
Some key features:
- Connect to tools in TypeScript and Go with (Python and .NET) (coming soon)
- Attach the tools to Agents (MCP / Open AI tool bindings)
- Long polling SDKs allow function calls beyond HTTP timeout limits and from within private VPCs
- Tracing, metrics, and events for visibility of every tool call
- Call tools from Claude Desktop / Cursor / any MCP client
Thanks taking a look, we would love your thoughts!
l1m.io
todai
Struggled this problem (RPC for AI) with the team in my current project. Go/Python also.
Will test it now. Seems very useful.
l1m.io
@goshatirov thanks for taking a look!
Reach out if you need anything / have any feedback 🙏
l1m.io
Hey hunters! Feel free to check out our examples for Typescript, Go and Python at https://github.com/agentrpc/agentrpc/tree/main/examples