mcp-test-stepfun
An MCP protocol test server with 35 test tools across 3 transport modes for compatibility verification, plus a development Skill for the Stepfun AI desktop assistant to help write MCP servers.
README
MCP Test Server
MCP 协议测试服务器 + 阶跃 AI 桌面助手开发 Skill。
<p> <a href="https://github.com/AIddlx/mcp-test-stepfun/stargazers"> <img src="https://img.shields.io/github/stars/AIddlx/mcp-test-stepfun?style=social" alt="GitHub stars"> </a> <a href="https://github.com/AIddlx/mcp-test-stepfun/network/members"> <img src="https://img.shields.io/github/forks/AIddlx/mcp-test-stepfun?style=social" alt="GitHub forks"> </a> </p>
项目价值
本项目提供两份核心资产:
| 资产 | 价值 |
|---|---|
| 兼容性测试 | 35 个测试工具 + 3 种传输模式,验证阶跃客户端 MCP 兼容性 |
| 开发 Skill | 6 条致命规则 + 17 条约束 + 3 套模板,让 AI 帮你写 MCP 服务器 |
一、兼容性测试
测试覆盖
| 类别 | 数量 | 说明 |
|---|---|---|
| A - 核心能力 | 5 | 协议版本、能力声明、工具调用、全类型参数 |
| B - 复杂参数 | 6 | 嵌套对象、大数据、长操作、并发、Unicode |
| C - 高级能力 | 4 | 进度通知、取消、批量请求、自动补全 |
| D - 边界条件 | 8 | 空参数、超长字符串、特殊字符、幂等性 |
| E - 极端条件 | 1 | 超时边界 |
| G - GUI Agent | 7 | 截图、鼠标、键盘、自动化演示 |
| H - Elicitation | 2 | 表单式、URL 式 |
| I - Sampling | 2 | 基础、带工具 |
传输模式
| 模式 | 目录 | 状态 |
|---|---|---|
| NPX (stdio) | stdio/npx/ |
35/35 通过 |
| UVX (stdio) | stdio/uvx/ |
35/35 通过 |
| HTTP/SSE | http/ |
35/35 通过 |
快速开始
HTTP 模式(推荐开发调试):
cd http
python full_test_server.py
阶跃客户端配置:
{
"mcpServers": {
"mcp-http-test": {
"url": "http://127.0.0.1:3372/mcp"
}
}
}
NPX 模式(推荐 stdio 测试):
{
"mcpServers": {
"mcp-npx-test": {
"command": "npx",
"args": ["-y", "<项目路径>/stdio/npx"]
}
}
}
UVX 模式(不推荐开发时用):
{
"mcpServers": {
"mcp-uvx-test": {
"command": "uvx",
"args": ["--from", "<项目路径>/stdio/uvx", "mcp-uvx-test"]
}
}
}
⚠️ UVX 模式有严重的缓存问题,每次修改源码都需要清理缓存 + 更新版本号。详见 docs/stdio/uvx/ISSUES.md。
二、开发 Skill
什么是 Skill
Skill 是阶跃 AI 桌面助手的扩展能力包。加载后,AI 获得"编写阶跃兼容 MCP 服务器"的专业知识。
Skill 内容
| 内容 | 说明 |
|---|---|
| 致命规则 TOP 6 | 违反必崩(outputSchema、async main、命令格式...) |
| 17 条约束 | 设计/编码/部署/排查全覆盖 |
| 决策树 | Connection closed 5 种原因排查 |
| 3 套模板 | NPX / UVX / HTTP 完整项目模板 |
使用方法
- 打包
skills/stepfun-mcp/文件夹为 zip - 在阶跃客户端中添加技能(导入 zip)
- 告诉阶跃你想实现的功能
示例对话:
"帮我写一个阶跃兼容的 MCP 服务器,提供文件读写功能"
项目结构
mcp-test-stepfun/
├── stdio/ # stdio 传输模式
│ ├── npx/ # Node.js (NPX)
│ └── uvx/ # Python (UVX)
├── http/ # HTTP/SSE 传输模式
├── sdk/ # SDK 测试服务器
│ ├── npx/ # TypeScript SDK
│ ├── http/ # Python SDK (HTTP)
│ └── uvx/ # Python SDK (stdio)
├── skills/stepfun-mcp/ # 开发 Skill
│ ├── SKILL.md # 主文件
│ ├── references/ # 参考文档
│ └── assets/ # 项目模板
└── docs/ # 技术文档
文档索引
必读
| 文档 | 说明 |
|---|---|
| docs/stdio/uvx/ISSUES.md | UVX 问题汇总(五层缓存) |
技术分析
| 文档 | 说明 |
|---|---|
| docs/FASTMCP_STREAMING_ANALYSIS.md | FastMCP "流式"原理 |
| docs/SSE_PROGRESS_DESIGN_NOTES.md | SDK 路由 Bug 分析 |
| docs/STEPFUN_STREAMING_LIMITATION.md | 阶跃流式限制 |
测试报告
| 文档 | 说明 |
|---|---|
| docs/stdio/npx/TEST_REPORT.md | NPX 模式测试报告 |
| docs/stdio/uvx/TEST_REPORT.md | UVX 模式测试报告 |
测试环境
- 平台: Windows 11 Pro
- 客户端: 阶跃 AI 桌面助手 v0.2.13
- 协议版本: MCP 2025-11-25
其他平台未测试。
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.