mcp-forge

mcp-forge

Turn any OpenAPI spec into a working MCP server — point it at a spec and Claude instantly gets a tool for every endpoint.

Category
访问服务器

README

mcp-forge

Turn any OpenAPI spec into a working MCP server. Point it at a spec (file or URL) and Claude instantly gets a tool for every endpoint — no codegen, no boilerplate, no hand-writing tool definitions.

npx mcp-forge ./openapi.json
# -> serving 14 tools from ./openapi.json -> https://api.example.com/v1

MCP is how Claude calls real tools. Writing an MCP server for an API you already have is tedious busywork — mcp-forge does it in one command by reading the API's own OpenAPI/Swagger spec at runtime and proxying the calls.

  • 🔌 Any REST API with an OpenAPI 3 / Swagger spec → Claude tools
  • Runtime, zero codegen — no generated files to maintain
  • 🧭 One tool per operation, typed inputs built from the spec's parameters
  • 🔐 Pass auth headers through to the upstream API
  • 📄 JSON or YAML specs, local file or URL

Use it with Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "petstore": {
      "command": "npx",
      "args": ["-y", "mcp-forge", "https://petstore3.swagger.io/api/v3/openapi.json"]
    }
  }
}

With auth + an explicit base URL:

{
  "mcpServers": {
    "myapi": {
      "command": "npx",
      "args": [
        "-y", "mcp-forge", "./openapi.yaml",
        "--base-url", "https://api.example.com/v1",
        "--header", "Authorization: Bearer YOUR_TOKEN"
      ]
    }
  }
}

CLI

mcp-forge <openapi-spec-file-or-url> [options]

  --base-url URL        Override the server URL (else taken from spec.servers[0])
  --header "Key: Value"  Header added to every upstream request (repeatable)
  --name NAME           MCP server name

As a library

import { loadSpec, openapiToTools, buildServer } from 'mcp-forge'
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js'

const spec = await loadSpec('./openapi.json')
const tools = openapiToTools(spec) // -> ToolDef[]
const server = buildServer(tools, { baseUrl: 'https://api.example.com/v1' })
await server.connect(new StdioServerTransport())

How it works

  1. Load + parse the spec (JSON/YAML, local or remote).
  2. Walk paths × HTTP methods → one tool per operation. The tool's input schema is built from the operation's path, query, and header parameters (plus a body for request bodies).
  3. When Claude calls a tool, mcp-forge builds the real HTTP request — filling path/query/header params and the JSON body — sends it, and returns the status + response.

Scope (v1)

Supports OpenAPI 3 operations with path/query/header params and JSON request bodies, and local #/... $ref resolution. Deep/remote $refs, oneOf/allOf body schemas, and non-JSON bodies are simplified — PRs welcome.

License

MIT © 2026 Neil Gilani

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选