Sezzlee Local LLM MCP server

Sezzlee MCP

Swagger for agents. Sezzlee turns what you already have — an HTTP backend, a folder of spreadsheets, a SQL Server or PostgreSQL database — into tools an AI agent can find and call over the Model Context Protocol.

It comes in two shapes:

search-first tool catalog and replays every agent call through your backend's own request pipeline, so your authentication, authorization and validation apply unchanged. For a backend on any other stack, an OpenAPI gateway builds the same catalog from its document.

• The HTTP catalog embeds in an ASP.NET Core or NestJS backend. It exposes your endpoints as a

XML, PDF, Microsoft SQL Server and PostgreSQL, plus a server that hands bounded language work to a local model.

• Source servers are standalone, read-only MCP servers for local files and databases: Excel,

Status: early. Every package below is on npm or NuGet; APIs may still change. The spec, the

fixture corpus and both SDKs are tested in CI.

Why

A naive MCP adapter turns every endpoint into a tool and forwards calls over the network. That floods the agent's context, loses the caller's identity and silently papers over what it cannot represent. Sezzlee instead:

whole catalog, so the size of your API does not grow the agent's context;

• shows the agent three meta-tools (searchtools, loadtool, invoke_tool) rather than the

call in the backend at invoke time;

• decides what a caller can see from the backend's own authorization, and still enforces every

an error with a code the agent can act on.

• treats every silent resolution — a name collision, an unknown argument, a truncated response — as

The longer argument is on the docs site: why sezzlee is not an OpenAPI adapter.

Quick start

Pick what you need. The servers need Node.js 22+; the ASP.NET Core SDK targets .NET 8 and 10.

Each package README has its own quick start, configuration and limits. To build from source, see CONTRIBUTING.md.

Packages

HTTP catalog

Shortened. The full README is on GitHub.