G3 General Purpose AI Agent
A code-first AI agent that helps you complete tasks by writing and executing code or scripts.
Philosophy
G3 doesn't just give you advice - it writes and runs code to solve your problems. Whether you need to:
- Process data files
- Automate workflows
- Scrape websites
- Manipulate files
- Analyze logs
- Generate reports
- Set up environments
G3 will write the appropriate scripts (Python, Bash, JavaScript, etc.) and can execute them for you.
Features
- Code-First Approach: Always tries to solve problems with executable code
- Multi-Language Support: Generates Python, Bash, JavaScript, Rust, and more
- Modular Architecture: Clean separation between CLI, core engine, and LLM providers
- Multiple LLM Providers: Support for OpenAI, Anthropic, and embedded open-weights models
- Local Model Support: Run completely offline with embedded GGUF models via llama.cpp
- Interactive Mode: Chat with the AI and watch it solve problems in real-time
- Task Automation: Create reusable automation scripts
Installation
cargo install --path .
Configuration
Create a configuration file at ~/.config/g3/config.toml:
Cloud Providers
[providers]
default_provider = "anthropic"
[providers.anthropic]
api_key = "your-anthropic-api-key"
model = "claude-3-5-sonnet-20241022"
max_tokens = 4096
temperature = 0.1
[providers.openai]
api_key = "your-openai-api-key"
model = "gpt-4"
max_tokens = 2048
temperature = 0.1
Local Embedded Models
For completely offline operation with open-weights models:
[providers]
default_provider = "embedded"
[providers.embedded]
# Path to your GGUF model file
model_path = "~/.cache/g3/models/codellama-7b-instruct.Q4_K_M.gguf"
model_type = "codellama"
context_length = 4096
max_tokens = 2048
temperature = 0.1
# Number of layers to offload to GPU (0 for CPU only)
gpu_layers = 32
# Number of CPU threads to use
threads = 8
Getting Models: Download GGUF models from Hugging Face (search for "GGUF"). Popular options:
Agent Settings
[agent]
max_context_length = 8192
enable_streaming = true
timeout_seconds = 60
Usage
Interactive Mode (Default)
Simply run G3 to start interactive mode:
g3
Example interactions:
g3> Process this CSV file and show me the top 10 customers by revenue
# G3 writes Python pandas script to analyze the CSV
g3> Set up a backup script for my home directory
# G3 creates a bash script with rsync/tar commands
g3> Download all images from this webpage
# G3 writes a Python script with requests/BeautifulSoup
g3> Convert these JSON files to a SQLite database
# G3 creates a Python script to parse JSON and insert into SQLite
Direct Commands
You can also use G3 with direct commands:
# Solve any task with code
g3 task "merge these PDF files into one"
g3 task "find all TODO comments in my codebase"
# Create automation scripts
g3 automate "daily backup of my projects folder"
g3 automate "resize all images in a folder to 800px width"
# Data processing
g3 data "analyze this log file for error patterns"
g3 data "convert CSV to JSON with validation"
# Legacy code commands (still supported)
g3 analyze src/main.rs
g3 generate "fibonacci function" --output fib.py
g3 review src/lib.rs
Architecture
G3 follows the Unix philosophy with modular, composable components:
- g3-cli: Command-line interface
- g3-core: Core agent logic and orchestration
- g3-providers: LLM provider abstractions
- g3-config: Configuration management
See DESIGN.md for detailed architecture documentation.
Development
# Build all crates
cargo build
# Run tests
cargo test
# Run with debug logging
RUST_LOG=debug cargo run -- analyze src/
License
MIT