# doc_agent

**Repository Path**: yujinlazyf/doc_agent

## Basic Information

- **Project Name**: doc_agent
- **Description**: 本次开发本地文档 + 天气综合 AI 智能体，基于 MCP 模型上下文协议实现工具调度，内置 3 个独立可调用工具：
read_file 工具：
write_summary 工具
query_weather 工具
- **Primary Language**: Python
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-06-20
- **Last Updated**: 2026-06-20

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Doc Agent

An intelligent document processing agent powered by LLM with MCP (Model Context Protocol) support.

## Overview

Doc Agent is an AI-powered document agent system that combines LLM decision-making capabilities with file operations and weather API tools. It uses the Model Context Protocol to orchestrate various tools and provide intelligent responses to user queries.

## Features

- **LLM-Powered Decision Making**: Uses large language models to analyze queries and decide appropriate actions
- **File Operations**: Read and write files with summary generation capabilities
- **Weather API Integration**: Query weather information by city name or geographic coordinates
- **MCP Server**: Implements Model Context Protocol for tool orchestration

## Project Structure

```
doc_agent/
├── agent_main.py       # Main agent logic with LLM decision functions
├── mcp_server.py       # MCP server implementation and tool definitions
├── tools/
│   ├── file_reader.py  # File reading functionality
│   ├── file_writer.py  # File writing and summary generation
│   └── weather_api.py  # Weather query API (supports China and international)
├── system_prompt.md    # System prompts for LLM
└── doc_summary.txt     # Document summaries storage
```

## Installation

```bash
# Install dependencies
pip install -r requirements.txt
```

## Usage

### Starting the Agent

```python
from agent_main import agent_run

# Run the agent
agent_run()
```

### Using MCP Tools

```python
from mcp_server import mcp_call_tool, list_available_tools

# List available tools
tools = list_available_tools()
print(tools)

# Call a specific tool
result = mcp_call_tool("read_file", {"file_path": "example.txt"})
```

### Tool Examples

**Read File:**
```python
from tools.file_reader import read_file

content = read_file("path/to/file.txt")
```

**Write Summary:**
```python
from tools.file_writer import write_summary

write_summary("output.txt", "Summary content here")
```

**Query Weather:**
```python
from tools.weather_api import query_cn_weather, query_weather

# Query by city name (China)
weather = query_cn_weather("北京")

# Query by coordinates
weather = query_weather(city_lat=39.9042, city_lon=116.4074)
```

## Available Tools

| Tool | Description |
|------|-------------|
| read_file | Read content from a specified file |
| write_summary | Write summary content to a file |
| query_cn_weather | Query weather by Chinese city name |
| query_weather | Query weather by latitude and longitude |

## Dependencies

- Python 3.10+
- pydantic (for data validation)
- requests (for API calls)
- llm SDK (for LLM integration)

## License

MIT License

## Contributing

Contributions are welcome! Please submit pull requests or open issues for any improvements.