Version 2.0

GitArsenal V2 Documentation

AI-powered autonomous agents that handle your entire development workflow—from repository analysis to deployment—in isolated cloud environments.

What's New in V2

🤖
Autonomous AI Agents

Intelligent agents analyze and set up your entire repository automatically

🐳
Multi-Container Orchestration

Automatic detection and provisioning of all required services

🔍
Smart Repository Analysis

Deep codebase understanding with framework and dependency detection

One-Command Setup

Single command installation with automatic environment configuration

CLI Commands & Options

All Commands

GitArsenal V2 provides a comprehensive set of commands for managing repositories, agents, and workspaces:

Main Commands
gitarsenal [project]           Start GitArsenal TUI (default command)
gitarsenal acp                  Start ACP (Agent Client Protocol) server
gitarsenal attach <url>          Attach to a running GitArsenal server
gitarsenal workspace <action>   Manage GPU workspaces
gitarsenal run [message..]       Run GitArsenal with a message
gitarsenal auth                  Manage credentials
gitarsenal agent                 Manage agents
gitarsenal upgrade [target]      Upgrade GitArsenal to latest or specific version
gitarsenal serve                 Start a headless GitArsenal server
gitarsenal web                   Start a headless opencode server
gitarsenal models [provider]    List all available models
gitarsenal stats                 Show token usage and cost statistics
gitarsenal export [sessionID]    Export session data as JSON
gitarsenal import <file>         Import session data from JSON file or URL
gitarsenal github                Manage GitHub agent
gitarsenal pr <number>           Fetch and checkout a GitHub PR branch, then run opencode

All Options

Complete list of all command-line options available in GitArsenal:

Global Options
# Help and Version
-h, --help              Show help message
-v, --version           Show version number

# Logging Options
--print-logs            Print logs to stderr
--log-level <level>     Set log level (DEBUG, INFO, WARN, ERROR)

# Model and Agent Options
-m, --model <provider/model>    Model to use (e.g., openai/gpt-4)
--agent <name>                  Agent to use

# Session Management
-c, --continue                  Continue the last session
-s, --session <id>              Session ID to continue
-p, --prompt <text>             Prompt to use

# Server Configuration
--port <number>                 Port to listen on (default: 0)
--hostname <address>             Hostname to listen on (default: 127.0.0.1)
--server <mode>                  Server mode: local, modal, or remote (default: local)
--url <address>                  Remote server URL when --server=remote (default: 127.0.0.1)

# Positional Arguments
<project>                       Path to start GitArsenal in

Server Modes:

local - Run on your local machine (default)
modal - Run in cloud workspace with GPU support
remote - Connect to an existing GitArsenal server URL

Usage Examples

Basic Usage
# Start interactive TUI (default)
gitarsenal

# Start in a specific project directory
gitarsenal /path/to/project

# Run with a specific message
gitarsenal run "Set up this repository with GPU support"

# Continue previous session
gitarsenal --continue

# Use specific model
gitarsenal --model openai/gpt-4

# Start in server mode
gitarsenal serve --port 8080

# Attach to remote server
gitarsenal attach https://remote-server.com

# Manage workspaces
gitarsenal workspace list
gitarsenal workspace create my-workspace

# Export session data
gitarsenal export session-123

# Import session data
gitarsenal import session-data.json

# Check GitHub PR
gitarsenal pr 42

Core Features

AI-Powered Setup

Autonomous agents analyze your repository structure, detect frameworks, identify dependencies, and automatically configure everything needed to run your code.

GPU Acceleration

Access powerful GPUs (T4, A10G, A100, H100) for machine learning, AI training, and compute-intensive workloads with automatic provisioning.

Cloud Isolation

Run your code safely in isolated cloud containers. Perfect for testing untrusted code, running experiments, or trying new frameworks.

Multi-Container Orchestration

Automatically spin up databases, caches, message queues, and all required services. The agent understands your architecture and provisions accordingly.

Persistent Storage

Keep your work, models, and data safe with persistent volumes that survive between sessions. Your progress is never lost.

IDE Integration

Use your favorite development tools: Claude Code, Gemini CLI, Cursor, or any SSH-capable editor. Complete control over your environment.

How It Works

1

Repository Analysis

The AI agent analyzes your repository structure, detects programming languages, frameworks, and dependencies. It understands your project's architecture and requirements.

2

Environment Provisioning

Based on the analysis, the agent provisions the exact containers, services, and resources needed. This includes databases, caches, GPU acceleration, and all dependencies.

3

Automatic Configuration

The agent configures everything automatically—environment variables, API keys, database connections, and service linking. No manual setup required.

4

Ready to Code

Connect with SSH or your preferred development tool. The environment is fully set up, dependencies installed, and services running. Start coding immediately.

What the AI Agent Can Do

Detection & Analysis

  • Detect programming languages and frameworks (Python, Node.js, Go, Rust, etc.)
  • Identify dependencies from requirements.txt, package.json, go.mod, etc.
  • Analyze Docker configurations and docker-compose.yml files
  • Detect required services (PostgreSQL, Redis, MongoDB, etc.)
  • Identify GPU requirements for ML/AI workloads

Provisioning & Setup

  • Provision compute resources and GPU acceleration
  • Spin up required databases and services
  • Install all dependencies automatically
  • Configure environment variables and secrets
  • Set up networking and service connections

Perfect For

🤖 AI & Machine Learning

Train models with GPU acceleration, experiment with different architectures, and deploy inference endpoints—all automatically configured.

🧪 Testing & Experimentation

Safely test untrusted code, try new frameworks, or experiment with different configurations in isolated environments.

🚀 Quick Prototypes

Clone any repository and have it running in minutes. Perfect for hackathons, demos, and rapid prototyping.

👥 Team Collaboration

Share consistent development environments across your team. Everyone gets the same setup, eliminating "works on my machine" issues.

📚 Learning & Education

Explore open-source projects, learn new frameworks, or teach students with ready-to-run environments.

🔬 Research & Development

Reproduce research experiments, run computationally intensive simulations, and collaborate on scientific projects.

V1 vs V2

FeatureV1V2
Repository AnalysisManual configuration✓ AI-powered automatic
Multi-Container SupportSingle container only✓ Full orchestration
Dependency DetectionManual specification✓ Automatic detection
Service ProvisioningNot supported✓ Databases, caches, etc.
Setup Time5-15 minutes✓ 2-5 minutes
Installationnpm/yarn packages✓ One-command installer

Ready to Get Started?

Install GitArsenal V2 and let AI agents handle your development workflow