SJTU MCP
Wraps the SJTU Zhiyuan No.1 API as a local MCP server, allowing direct use of SJTU models in Claude Code and Codex without custom integration scripts.
README
SJTU MCP
English | 中文
Turn your SJTU Zhiyuan No.1 API key into something you can actually use in Claude Code and Codex.
SJTU MCP wraps the SJTU-hosted model API as a local MCP server, so you can call these models directly from your normal agent workflow instead of hand-writing integration scripts over and over again.
Why This Exists
Have you already applied for an SJTU Zhiyuan No.1 API key, but still found it hard to actually use in practice?
This project exists to solve exactly that problem:
- you already have API access
- you want to use it from
Claude CodeorCodex - but the SJTU endpoint itself does not plug directly into these agent tools out of the box
- you do not want to rewrite the integration layer every time
Highlights
- Supports
Claude Code - Supports
Codex - Supports both text and vision tasks
- Uses the SJTU OpenAI-compatible endpoint
- Fits naturally into existing MCP workflows
Contents
Quick Start
For most users, the simplest path is:
git clonethis repocdinto the project directory- install it once
- add it as a global MCP server in
Claude CodeorCodex
git clone https://github.com/EternalWavee/sjtu-mcp.git
cd sjtu-mcp
pip install -e .
After installation, your MCP client can start the server automatically when needed. In normal use, you do not need to manually run the server command every time.
Environment Variables
Required:
SJTU_API_KEY
Optional:
SJTU_API_BASE_URLSJTU_DEFAULT_TEXT_MODELSJTU_DEFAULT_REASONING_MODELSJTU_DEFAULT_VISION_MODELSJTU_REQUEST_TIMEOUT
How to use them:
.env.exampleis only a template showing which variables you need- in actual use, put these values into the
envblock of your MCP configuration
Claude Code
Recommended: User Scope
Use this if you want sjtu available in all your Claude Code projects on this machine.
claude mcp add sjtu --scope user -- python -m sjtu_mcp.server
Then:
- open
~/.claude.json - find the
sjtuentry - copy the
envsection from examples/claude-project.mcp.json - replace
your-api-keywith your real key
Verify:
claude mcp list
Project Scope
Use this if you want to commit a shared config into the repo for teammates.
How to use it:
- copy examples/claude-project.mcp.json into your project root as
.mcp.json - replace
your-api-keywith your real key - adjust default models and timeout if needed
Windows / macOS example:
{
"mcpServers": {
"sjtu": {
"command": "python",
"args": ["-m", "sjtu_mcp.server"],
"env": {
"SJTU_API_BASE_URL": "https://models.sjtu.edu.cn/api/v1",
"SJTU_API_KEY": "your-api-key",
"SJTU_DEFAULT_TEXT_MODEL": "deepseek-chat",
"SJTU_DEFAULT_REASONING_MODEL": "deepseek-reasoner",
"SJTU_DEFAULT_VISION_MODEL": "qwen3vl",
"SJTU_REQUEST_TIMEOUT": "180"
}
}
}
}
Local Scope
Use this if you only want the server for the current project and do not want to commit the config.
claude mcp add sjtu --scope local -- python -m sjtu_mcp.server
Then add the same env values to the corresponding MCP config entry.
Codex
Recommended: Global Setup
Use this if you want sjtu available in all your Codex projects on this machine.
codex mcp add sjtu -- python -m sjtu_mcp.server
Then:
- open your own
~/.codex/config.toml - copy the content from examples/codex-config.toml
- replace
your-api-keywith your real key - save and reload Codex or reload MCP
Verify:
codex mcp list
Config File Setup
If you already manage ~/.codex/config.toml directly, you can use this template:
[mcp_servers.sjtu]
command = "python"
args = ["-m", "sjtu_mcp.server"]
[mcp_servers.sjtu.env]
SJTU_API_BASE_URL = "https://models.sjtu.edu.cn/api/v1"
SJTU_API_KEY = "your-api-key"
SJTU_DEFAULT_TEXT_MODEL = "deepseek-chat"
SJTU_DEFAULT_REASONING_MODEL = "deepseek-reasoner"
SJTU_DEFAULT_VISION_MODEL = "qwen3vl"
SJTU_REQUEST_TIMEOUT = "180"
Tools
sjtu_modelssjtu_textsjtu_visionsjtu_cheap_task
Example
Input
请调用 sjtu_vision 分析图片里面的内容 .assets/test.png

Output

Suggested Model Usage
deepseek-chat- default for summaries, rewrites, cleanup, and low-risk text tasks
minimaxorglm-5- useful for lightweight rewriting, classification, or extraction
deepseek-reasoner- better for tasks that truly need multi-step reasoning
qwen3vl- a strong starting point for screenshots, OCR-style extraction, and image understanding
qwen3coder- useful for code-adjacent utility tasks
Notes
- This server currently assumes the SJTU endpoint supports OpenAI-compatible
/modelsand/chat/completions. - Local images are encoded as data URLs before sending.
- If your campus endpoint has model-specific quirks, extend the routing in src/sjtu_mcp/server.py.
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.