Back to packageRELEASE DETAILS

0.1.9

PyPI

agimat-agent

基于 FastAPI 和 LangGraph 的智能代理管理平台

Published12 Jul 2025

Direct dependencies45

Transitive dependenciesNot calculated

README

Version 0.1.9 documentation

🤖 AGI-MAT Agent Management Platform

基于 FastAPI 和 LangGraph 的智能代理管理平台，支持多种大语言模型和MCP工具集成。

✨ 主要特性

🎯 多模型支持 - 支持 Claude、Gemini、GPT、豆包等主流模型
🛠️ MCP工具集成 - 内置本地工具和外部MCP服务器支持
🔄 流式响应 - 支持实时流式对话和WebSocket连接
📊 可观测性 - 集成LangSmith追踪和监控
🔌 插件化架构 - 灵活的工具和模型扩展机制

🚀 支持的模型

Anthropic Claude

claude-3-5-sonnet - 最新平衡型模型
claude-3-5-haiku - 快速轻量级模型
claude-3-opus - 最强推理能力模型

Google Gemini

gemini-1.5-pro - 高性能多模态模型
gemini-1.5-flash - 快速响应模型

OpenAI

gpt-4o-mini - 轻量级多模态模型
gpt-4o - 完整多模态模型

国产模型

doubao-1-5-lite-32k-250115 - 字节跳动豆包

📦 快速开始

1. 环境准备

# 克隆项目
git clone <repository-url>
cd agimat-agent

# 创建虚拟环境
python -m venv .venv
source .venv/bin/activate  # Linux/Mac
# 或 .venv\Scripts\activate  # Windows

# 安装依赖
pip install -r requirements.txt

2. 使用 uv 管理环境依赖

# 同步依赖
uv sync

# 直接启动
uv run main.py

🖥️ CLI 命令行工具

AGI-MAT Agent 提供了便捷的命令行工具来管理服务。

基本命令

# 查看帮助
agimat --help

# 启动后端服务器
agimat server

# 启动前端UI
agimat ui

# 打开配置目录
agimat config

详细用法

启动后端服务器

# 使用默认配置启动服务器
agimat server

# 指定端口启动
agimat server --port 9000

# 指定主机地址
agimat server --host 0.0.0.0 --port 8000

# 使用自定义配置文件
agimat server --config /path/to/custom/config.yaml

启动前端UI

# 使用默认端口启动UI
agimat ui

# 指定UI端口
agimat ui --port 8502

# 使用自定义配置文件
agimat ui --config /path/to/custom/config.yaml

管理配置文件

# 打开配置目录（使用系统默认方式）
agimat config

# 使用VS Code打开配置目录
agimat config --editor code

# 使用vim打开配置目录
agimat config --editor vim

# 使用nano打开配置目录
agimat config --editor nano

配置文件说明

AGI-MAT Agent 会自动在 ~/.agimat/ 目录下创建配置文件：

主配置文件 (`~/.agimat/conf.yaml`)

包含 AGI-MAT Agent 的基本配置，如服务器地址、模型参数等。

环境变量文件 (`~/.agimat/env`)

包含各种API密钥和服务器配置，如：

OpenAI API 配置
Google AI 配置
Anthropic 配置
LangSmith 配置
数据库配置
服务器配置
日志配置

MCP配置文件 (`~/.agimat/mcp.json`)

包含MCP服务器的配置信息，支持各种外部工具集成。

配置优先级

命令行参数指定的配置文件
~/.agimat/conf.yaml
项目根目录的 conf.yaml
默认配置

首次运行

首次运行 agimat 命令时，系统会自动：

创建 ~/.agimat/ 配置目录
生成默认的配置文件
创建环境变量模板文件
设置MCP工具配置

故障排除

常见问题

Q: 配置文件在哪里？ A: 默认位置是 ~/.agimat/ 目录，使用 agimat config 命令可以快速打开。

Q: 如何修改API密钥？ A: 编辑 ~/.agimat/env 文件，或者使用 agimat config --editor code 打开配置目录。

Q: 端口被占用怎么办？ A: 使用 --port 参数指定其他端口，例如：agimat server --port 9000

Q: 如何查看日志？ A: 在 ~/.agimat/env 文件中设置 LOG_LEVEL=DEBUG 查看详细日志。

3. 配置模型

使用CLI工具快速配置：

# 打开配置目录
agimat config

# 或者使用VS Code编辑配置
agimat config --editor code

编辑 ~/.agimat/env 文件，配置你的API密钥：

# OpenAI API 配置
OPENAI_API_KEY=your_openai_api_key_here

# Google AI 配置  
GOOGLE_API_KEY=your_google_api_key_here

# Anthropic 配置
ANTHROPIC_API_KEY=your_anthropic_api_key_here

4. 启动服务

使用CLI工具启动服务：

# 启动后端服务器
agimat server

# 启动前端UI
agimat ui

# 或者同时启动两个服务（在不同终端）
agimat server --port 8000
agimat ui --port 8501

服务将在 http://localhost:8000 启动，UI在 http://localhost:8501。

🔧 使用示例

API调用

# 基础对话
curl -X POST "http://localhost:8000/api/v1/agents/invoke" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_name": "agent_test",
    "query": "你好，请介绍一下你自己",
    "task_id": "test-task-1"
  }'

流式对话

# SSE流式响应
curl -X POST "http://localhost:8000/api/v1/agents/stream" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_name": "agent_test", 
    "query": "写一首关于AI的诗",
    "task_id": "stream-task-1"
  }'

WebSocket连接

const ws = new WebSocket('ws://localhost:8000/api/v1/agents/ws');

ws.onopen = () => {
  ws.send(JSON.stringify({
    agent_name: "agent_test",
    query: "解释一下量子计算的原理",
    task_id: "ws-task-1"
  }));
};

ws.onmessage = (event) => {
  const data = JSON.parse(event.data);
  console.log(data);
};

🧪 测试配置

运行模型配置测试：

# 完整测试
python test_models.py

# 快速测试
python test_models.py --quick

📚 文档

多模型配置指南 - 详细的模型配置说明
MCP工具文档 - MCP工具集成指南
API文档 - 启动服务后访问

🏗️ 项目结构

agimat-agent/
├── api/                    # API路由层
│   └── agent.py           # Agent API端点
├── core/                  # 核心业务逻辑
│   ├── apps/
│   │   ├── agents/        # Agent实现
│   │   ├── mcp/          # MCP工具集成
│   │   ├── tools/        # 本地工具
│   │   └── builder/      # Agent构建器
│   └── models/           # 数据模型
├── lib/                  # 通用库
│   ├── llm_provider.py   # 多模型支持
│   └── config.py         # 配置管理
├── conf.yaml            # 主配置文件
├── main.py             # 应用入口
└── requirements.txt    # 依赖包

🎨 UI 启动

使用CLI工具启动UI：

# 使用默认端口启动UI
agimat ui

# 指定端口启动UI
agimat ui --port 8501

# 使用uv直接启动（如果需要）
uv run streamlit run ui/ui.py --server.port 8501

🤝 贡献指南

Fork 本仓库
创建特性分支 (git checkout -b feature/AmazingFeature)
提交更改 (git commit -m 'Add some AmazingFeature')
推送到分支 (git push origin feature/AmazingFeature)
创建 Pull Request

📄 许可证

本项目采用 MIT 许可证 - 查看 LICENSE 文件了解详情。

🙏 致谢

LangChain - 强大的LLM应用框架
FastAPI - 现代化的Web API框架
MCP - 模型上下文协议

Downloads & popularity

Awaiting observations

No metric observations yet

Download counts and popularity will appear here as the collector receives them.

Declared dependencies 45

Native requirements

45 direct targetsNot calculated transitive count

Package / target	Requirement	Relationship
aiofiles Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`aiofiles`	runtime
aiohttp Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`aiohttp`	runtime
build Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`build>=1.2.2.post1`	runtime
email-validator Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`email-validator`	runtime
fastapi Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`fastapi`	runtime
fastmcp Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`fastmcp`	runtime
google-generativeai Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`google-generativeai`	runtime
jinja2 Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`jinja2`	runtime
langchain Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain`	runtime
langchain-anthropic Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-anthropic`	runtime
langchain-community Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-community`	runtime
langchain-core Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-core`	runtime
langchain-experimental Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-experimental`	runtime
langchain-google-genai Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-google-genai`	runtime
langchain-mcp-adapters Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-mcp-adapters`	runtime
langchain-openai Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-openai`	runtime
langchain-tavily Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langchain-tavily`	runtime
langgraph Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langgraph`	runtime
langsmith Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`langsmith`	runtime
mcp Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`mcp`	runtime
motor Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`motor`	runtime
openai Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`openai`	runtime
passlib Environment { "extras": "bcrypt", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`passlib[bcrypt]`	runtime
psutil Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`psutil`	runtime
pydantic Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pydantic`	runtime
pydantic-settings Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pydantic-settings`	runtime
python-dateutil Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`python-dateutil`	runtime
python-dotenv Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`python-dotenv`	runtime
python-jose Environment { "extras": "cryptography", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`python-jose[cryptography]`	runtime
python-multipart Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`python-multipart`	runtime
pytz Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pytz`	runtime
pyyaml Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pyyaml`	runtime
requests Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`requests`	runtime
streamlit Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`streamlit>=1.28.0`	runtime
tenacity Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`tenacity`	runtime
twine Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`twine>=6.1.0`	runtime
uvicorn Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`uvicorn`	runtime
wavespeed-mcp Environment { "extras": "", "marker": null, "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`wavespeed-mcp>=0.1.6`	runtime
black Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`black>=24.1.1; extra == "dev"`	runtime
flake8 Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`flake8>=7.0.0; extra == "dev"`	runtime
httpx Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`httpx>=0.27; extra == "dev"`	runtime
isort Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`isort>=5.13.2; extra == "dev"`	runtime
mypy Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`mypy>=1.8.0; extra == "dev"`	runtime
pytest-asyncio Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pytest-asyncio>=0.23.5; extra == "dev"`	runtime
pytest Environment { "extras": "", "marker": "extra == \"dev\"", "requires_python": ">=3.12", "evaluation": "NOT_EVALUATED" }	`pytest>=8.0.0; extra == "dev"`	runtime

Distribution files

2 collected

agimat_agent-0.1.9-py3-none-any.whl

bdist_wheel · 62.67 KB · Python py3

Source file

agimat_agent-0.1.9.tar.gz

sdist · 39.37 KB · Python source