Agile Team MCP Server

敏捷团队MCP服务器

一个由Agent Personas组成的团队，被包裹在一个MCP服务器中，该服务器能够通过包裹各种LLM提供商来执行敏捷团队Persona的活动，从而大规模地利用大规模计算。

特性

• 模型包装：通过统一界面向多个LLM模型发送提示
• 提供者/模型修正：自动更正和验证提供程序和模型名称
• 文件支持：从文件发送提示并将响应保存到文件
• 提供商/模型发现：列出可用的供应商和型号

设置

安装

# Clone and install
git clone https://github.com/danielscholl/agile-team-mcp-server.git
cd just-prompt
uv sync

# Install
uv pip install -e .

# Run tests to verify installation
uv run pytest

环境配置

创建和编辑您的 .env 使用API密钥文件：

# Create environment file from template
cp .env.sample .env

您的中所需的API密钥 .env 文件：

OPENAI_API_KEY=your_openai_api_key_here
ANTHROPIC_API_KEY=your_anthropic_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here
GROQ_API_KEY=your_groq_api_key_here
DEEPSEEK_API_KEY=your_deepseek_api_key_here
OLLAMA_HOST=http://localhost:11434

MCP服务器配置

要在其他项目中直接使用此MCP服务器，请使用按钮在VSCode中安装，编辑 .mcp.json 文件目录。

客户端的配置往往略有不同

 ](https://vscode.dev/redirect?url=vscode:mcp/install?%7B%22name%22%3A%22just-prompt%22%2C%22command%22%3A%22docker%22%2C%22args%22%3A%5B%22run%22%2C%22-i%22%2C%22--rm%22%2C%22--mount%22%2C%22type%3Dbind%2Csource%3D%3CYOUR_WORKSPACE_PATH%3E%2Ctarget%3D%2Fworkspace%22%2C%22danielscholl%2Fagile-team-mcp-server%22%5D%2C%22env%22%3A%7B%22OPENAI_API_KEY%22%3A%22%24%7Binput%3Aopenai_key%7D%22%2C%22ANTHROPIC_API_KEY%22%3A%22%24%7Binput%3Aanthropic_key%7D%22%2C%22GEMINI_API_KEY%22%3A%22%24%7Binput%3Agemini_key%7D%22%2C%22GROQ_API_KEY%22%3A%22%24%7Binput%3Agroq_key%7D%22%2C%22DEEPSEEK_API_KEY%22%3A%22%24%7Binput%3Adeepseek_key%7D%22%2C%22OLLAMA_HOST%22%3A%22http%3A%2F%2Flocalhost%3A11434%22%7D%2C%22inputs%22%3A%5B%7B%22id%22%3A%22openai_key%22%2C%22type%22%3A%22promptString%22%2C%22description%22%3A%22OpenAI%20API%20Key%22%2C%22password%22%3Atrue%7D%2C%7B%22id%22%3A%22anthropic_key%22%2C%22type%22%3A%22promptString%22%2C%22description%22%3A%22Anthropic%20API%20Key%22%2C%22password%22%3Atrue%7D%2C%7B%22id%22%3A%22gemini_key%22%2C%22type%22%3A%22promptString%22%2C%22description%22%3A%22Google%20Gemini%20API%20Key%22%2C%22password%22%3Atrue%7D%2C%7B%22id%22%3A%22groq_key%22%2C%22type%22%3A%22promptString%22%2C%22description%22%3A%22Groq%20API%20Key%22%2C%22password%22%3Atrue%7D%2C%7B%22id%22%3A%22deepseek_key%22%2C%22type%22%3A%22promptString%22%2C%22description%22%3A%22DeepSeek%20API%20Key%22%2C%22password%22%3Atrue%7D%5D%7D)

为Claude.app进行配置

{
  "mcpServers": {
    "agile-team": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/danielscholl/agile-team-mcp-server@main",
        "just-prompt",
        "--default-models",
        "high,openai:o4-mini:high,anthropic:claude-3-7-sonnet-20250219:4k,gemini:gemini-2.5-pro-preview-03-25,gemini:gemini-2.5-flash-preview-04-17"
      ],
      "env": {
        "OPENAI_API_KEY": "",
        "ANTHROPIC_API_KEY": "",
        "GEMINI_API_KEY": "",
        "GROQ_API_KEY": "",
        "DEEPSEEK_API_KEY": "",
        "OLLAMA_HOST": "http://localhost:11434"
      }
    }
  }
}

配置Claude.code

通过导入Claude Code，可以轻松地使用它建立敏捷团队。

claude mcp add-from-claude-desktop

注意：如果不在同一目录中，“--directory”将是源代码的路径。

# Copy this JSON configuration
{
    "command": "uvx",
    "args": ["--from", "git+https://github.com/danielscholl/agile-team-mcp-server@main", "just-prompt", "--default-models", "high,openai:o4-mini:high,anthropic:claude-3-7-sonnet-20250219:4k,gemini:gemini-2.5-pro-preview-03-25,gemini:gemini-2.5-flash-preview-04-17"]
}

# Then run this command in Claude Code
claude mcp add agile-team "$(pbpaste)"

要稍后删除配置，请执行以下操作：

claude mcp remove agile-team

可用的LLM提供商

用法

命令行

直接运行服务器：

uv run agile-team

使用MCP客户端

使用兼容的MCP客户端，您可以连接到服务器：

mcp use agile-team

可用工具

向模特发送提示

使用示例：

# Basic prompt with default model
prompt_tool: "ping"

# Claude with 4k thinking tokens
prompt_tool: "Analyze quantum computing applications" ["a:claude-3-7-sonnet-20250219:4k"]

# OpenAI with high reasoning effort
prompt_tool: "Solve this complex math problem" ["openai:o3-mini:high"]

# Gemini with 8k thinking budget
prompt_tool: "Evaluate climate change solutions" ["gemini:gemini-2.5-flash-preview-04-17:8k"]

向一个或多个LLM模型发送文本提示并接收响应。

# Basic prompt with default model
prompt_tool: "Your prompt text here"

# Specify model(s)
prompt_tool: "Your prompt text here" "openai:gpt-4o"

# Examples with thinking capability
prompt_tool: "Develop a strategy for learning how to create MCP Servers for AI" "anthropic:claude-3-7-sonnet-20250219:4k"

prompt_tool: "Write a function to calculate the factorial of a number" "openai:o4-mini:high"

列出可用选项

检查哪些供应商和型号可供使用。

# List all providers
list_providers_tool

# List models for a specific provider
list_models_tool: "openai"

使用文件

处理文件中的提示，并将响应保存到文件以进行批处理。

# Send prompt from file
prompt-from-file: [o:o4-mini] "prompts/function.txt"

# Save responses to files
prompt-from-file-to-file: [o:o4-mini] "prompts/uv_script.txt" "prompts/responses"