mirror of https://github.com/hanzoai/mcp.git synced 2026-10-15 12:30:12 +00:00

Files

T

zeekay 96a9adc7aa Add Sphinx documentation

This commit adds a comprehensive Sphinx documentation system to the project:

- Created a new docs/ directory with Sphinx configuration
- Added documentation pages for installation, quick start, IDE integration
- Added security guide, debugging guide, and system prompt documentation
- Included guides for useful prompts and commands
- Set up the Read the Docs theme for better readability
- Added .gitignore for build files

This documentation system makes the project more accessible to new users
and provides centralized, browsable documentation that can be built into
HTML, PDF, or other formats as needed.

2025-04-12 23:47:37 -05:00

2.7 KiB

Raw Permalink Blame History

Quick Start Guide

This guide will help you get started with Hanzo MCP quickly.

Basic Usage

After installing Hanzo MCP, you can start using it with the following steps:

Start the MCP server:

hanzo-mcp --allow-path /path/to/your/project

Configure your AI assistant to use the MCP server.

Advanced Configuration

LLM Model and Token Configuration

You can customize the agent behavior by specifying the LLM model and token limits:

{
  "mcpServers": {
    "claude-code": {
      "command": "uvx",
      "args": [
        "--from",
        "hanzo-mcp",
        "hanzo-mcp",
        "--allow-path",
        "/path/to/project",
        "--enable-agent-tool",
        "--agent-model",
        "{litellm-model-name}",
        "--agent-max-tokens",
        "100000",
        "--agent-api-key",
        "{your-api-key}",
        "--agent-max-tool-uses",
        "30",
        "--agent-max-iterations",
        "10"
      ]
    }
  }
}

Available Configuration Options

The available LLM configuration options are:

--agent-model: Specify the model name in LiteLLM format (e.g., 'openai/gpt-4o', 'anthropic/claude-3-sonnet')
--agent-max-tokens: Specify the maximum tokens for agent responses
--agent-api-key: Specify the API key for the LLM provider (for development/testing only)
--agent-max-iterations: Maximum number of iterations for agent (default: 10)
--agent-max-tool-uses: Maximum number of total tool uses for agent (default: 30)
--enable-agent-tool: Enable the agent tool (disabled by default)

Model Name Formats

The model name uses the LiteLLM format with provider prefixes. Examples:

OpenAI models: openai/gpt-4o, openai/gpt-4o-mini
Anthropic models: anthropic/claude-3.7-sonnet
Google models: openrouter/google/gemini-2.0-flash-001 (Recommended)

Configuring System Prompt

To get the best experience with Hanzo, you should configure the system prompt in your AI assistant:

Locate the system prompt file in the repository at doc/system_prompt
Add it to your AI assistant's system prompt or project instructions
Replace {project_path} in the prompt with the absolute path to your project

The system prompt provides your AI assistant with:

A structured workflow for analyzing and modifying code
Best practices for project exploration and analysis
Guidelines for development, refactoring, and quality assurance
Special formatting instructions for mathematical content

Next Steps

Learn about IDE integration for using Hanzo MCP with your favorite editor
Check out debugging techniques for troubleshooting
Review security considerations for safe usage

2.7 KiB Raw Permalink Blame History