调用魔搭 Qwen3-VL 多模态模型识别作业图片中的题目与答案，逐题判分并给出解析，支持批量批改。
接入配置
选择客户端，复制配置，再按项目文档补齐环境变量。
claude mcp add pickstar-2002 --env MODELSCOPE_API_KEY='<MODELSCOPE_API_KEY>' -- npx @pickstar-2002/homework-grading-mcp@latest~/.codex/config.toml
[mcp_servers.pickstar-2002]
command = "npx"
args = ["@pickstar-2002/homework-grading-mcp@latest"]
env = { MODELSCOPE_API_KEY = "<MODELSCOPE_API_KEY>" }
# codex mcp add pickstar-2002 --env MODELSCOPE_API_KEY='<MODELSCOPE_API_KEY>' -- npx @pickstar-2002/homework-grading-mcp@latest~/.cursor/mcp.json
{
"mcpServers": {
"pickstar-2002": {
"command": "npx",
"args": [
"@pickstar-2002/homework-grading-mcp@latest"
],
"env": {
"MODELSCOPE_API_KEY": "<MODELSCOPE_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"pickstar-2002": {
"type": "stdio",
"command": "npx",
"args": [
"@pickstar-2002/homework-grading-mcp@latest"
],
"env": {
"MODELSCOPE_API_KEY": "<MODELSCOPE_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"pickstar-2002": {
"command": "npx",
"args": [
"@pickstar-2002/homework-grading-mcp@latest"
],
"env": {
"MODELSCOPE_API_KEY": "<MODELSCOPE_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"pickstar-2002": {
"command": "npx",
"args": [
"@pickstar-2002/homework-grading-mcp@latest"
],
"env": {
"MODELSCOPE_API_KEY": "<MODELSCOPE_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"pickstar-2002": {
"command": "npx",
"args": [
"@pickstar-2002/homework-grading-mcp@latest"
],
"env": {
"MODELSCOPE_API_KEY": "<MODELSCOPE_API_KEY>"
}
}
}
}需要的环境变量
MODELSCOPE_API_KEY
把占位值换成你自己的密钥；不要把密钥提交到代码仓库。