From REST API to MCP Tool in Minutes - No Code Required

Every team building with AI agents faces the same wall: your existing APIs, internal tools, and services aren't MCP-compatible, and making them so means writing and maintaining MCP server code for every integration. The result is fragmented tooling, duplicated effort, and agents that can't share context across sessions or clients.

What if you could point at any REST API and have it available as an MCP tool immediately, no code, no custom server, no redeployment?

In this session I'll demonstrate an open-source MCP gateway that does exactly this. Define an endpoint, map its parameters, and your tool is live, accessible to any MCP-compatible AI client instantly. Beyond tool creation, the gateway handles cross-session context sharing so agents on different clients can collaborate around shared state, and a skills layer lets you define reusable prompt workflows that any agent can invoke by name.

I'll close with a live demo: an audience member suggests any REST API on the spot - or we pick one at random, and within minutes it's a working MCP tool, callable from any AI client in the room. No code written, no server restarted.

• How to expose any REST API as an MCP tool without writing server code

• Patterns for cross-session context sharing across AI clients and frameworks

• How to build reusable skill libraries on top of MCP

• A live demo of agent-driven tool registration at runtime