This repository provides a hands-on workshop for building AI coding agents with progressively increasing capabilities, using Anthropic's Claude API and Go. It's designed for developers looking to understand agent architecture, tool integration, and practical implementation, offering a clear progression from a basic chat interface to agents capable of file system access, code search, and shell command execution.

How It Works

The project employs a common architecture centered around a robust event loop. This loop manages user input, orchestrates calls to the Anthropic Claude API, processes tool-use requests, and handles the execution and feedback of registered tools. Each agent implementation incrementally adds new tool capabilities, demonstrating concepts like schema generation, error handling, and progressive enhancement of agent functionality.

Quick Start & Requirements

• Primary install/run command: go run <agent_file.go> (e.g., go run chat.go)
• Prerequisites: Go 1.24.2+ or devenv (recommended), Anthropic API key (set as ANTHROPIC_API_KEY environment variable).
• Setup: devenv shell for a pre-configured environment, or go mod tidy for manual setup.
• Docs: Blog post available at https://ghuntley.com/agent/

Highlighted Details

• Six distinct agent implementations showcase a clear progression of capabilities, from basic chat to advanced code search.
• Demonstrates a consistent "Tool Definition Pattern" for integrating new functionalities.
• Utilizes automatic JSON schema generation from Go structs for tool parameter validation.
• Offers a --verbose flag for detailed logging of API calls, tool execution, and file operations.

Maintenance & Community

The repository appears to be a personal project by ghuntley, with no specific community links (Discord, Slack) or prominent contributors mentioned in the README.

Licensing & Compatibility

The repository's license is not explicitly stated in the provided README.

Limitations & Caveats

The README does not specify any limitations or known issues. The project relies on the Anthropic API, so access and costs associated with that service are implicit requirements.