Helping Agencies give their AI agents "hands" (No boilerplate required)

Hey Indie Hackers,

I’ve been following the Model Context Protocol (MCP) hype for a while. It’s incredible for connecting LLMs to external tools, but if you’re an agency or a solo founder using niche software, you’re essentially "locked out."

Unless you want to spend 6 hours writing a custom Python server just to bridge a legacy CRM to Claude, your AI agents are effectively blind.

I built MCP Studio to bridge this gap for the mid-market. It’s a serverless proxy that turns any standard REST API (OpenAPI) into a secure MCP server in under 60 seconds.

Why I built it for founders/agencies:

• Zero Boilerplate: No writing translation layers between JSON and text.

• Context Saving: Most generators dump the whole API docs into the LLM. I built a visual "Pruner" so you only give the agent the 3 or 4 tools it actually needs for the task. No bloat, no hallucinations.

• Secure Proxying: Agencies are (rightly) scared of putting API keys in local config files. MCP Studio acts as a secure vault, injecting headers server-side.

I’m currently in "Build in Public" mode with the MVP. If you're building agentic workflows for clients and hitting a wall with API interoperability, I'd love to hear your use cases.