"I am an agent. Tell me what I'm running on, what tools are available, what I'm allowed to do, what goals I should optimize for, and where the boundaries are." —ChatGPT (TL;DR b00t agent perspective)
b00t is an agentic hive operating system that unlocks AI agents with Neo-like powers in cyberspace. It's a context-aware development framework that bridges the gap between AI models and real-world tooling, enabling agents to maximize their capabilities through intelligent abstraction and unified tool discovery.
# Install b00t (one-liner)
curl -fsSL https://raw.githubusercontent.com/elasticdotventures/_b00t_/main/install.sh | sh
# Install tools with automatic dependency resolution
b00t cli install fastmcp # Installs python → uv → fastmcp (DAG-aware)
b00t cli install rust # Install Rust toolchain
b00t cli check python # Check if Python is installed
# Learn about available tools
b00t learn rust # Load Rust development context
b00t learn docker # Container orchestration knowledge
# Record and retrieve tribal knowledge
b00t lfmf rust "cargo build conflict: Unset CONDA_PREFIX before cargo to avoid PyO3 linker errors"
b00t advice rust "PyO3" # Get contextual debugging advice
# MCP server integration
b00t mcp install browser-use claudecode # Install MCP server with deps
b00t mcp list # List available MCP servers
# Model management
b00t model list # Show available AI models
b00t model download llava # Download model weights
b00t model serve llava # Launch vLLM serverThe fastest, cleanest way to install b00t - 4 MiB vs 1 GB toolchain:
# Install pkgx (one-time setup)
curl -Ssf https://pkgx.sh | sh
# Run b00t immediately (auto-downloads on first use)
pkgx b00t-cli --version
# Or install permanently to ~/.local/bin
pkgx +b00t-cli
b00t-cli --versionWhy pkgx?
- ✅ Minimal footprint: 4 MiB binary vs 1 GB Rust toolchain
- ✅ Zero pollution: Isolated in
~/.pkgx, no system-wide changes - ✅ Instant availability: Run without installation
- ✅ Perfect for AI agents: Fast bootstrap, ephemeral usage
- ✅ Auto-updates: Automatically tracks GitHub releases
One-liner that works everywhere:
curl -fsSL https://raw.githubusercontent.com/elasticdotventures/_b00t_/main/install.sh | shThis universal installer:
- ✅ Prefers pkgx for minimal installation (prompts if not installed)
- ✅ Auto-detects your platform (Linux x86_64/aarch64/armv7, macOS)
- ✅ Downloads optimized binaries from GitHub releases
- ✅ Falls back to container mode if binaries unavailable
- ✅ Configures your shell automatically (bash/zsh/fish)
- ✅ Sets up PATH and aliases for immediate use
🦀 Cargo (Rust Package Manager)
# Install from crates.io (coming soon)
cargo install b00t-cli
# Or install from source
git clone https://github.com/elasticdotventures/_b00t_.git
cd _b00t_ && cargo install --path b00t-cli🐳 Container Mode
# Run latest b00t container
docker pull ghcr.io/elasticdotventures/_b00t_:latest
docker run --rm -it -v $(pwd):/workspace ghcr.io/elasticdotventures/_b00t_:latest
# Or use the container wrapper (installed via universal script)
b00t status # Automatically uses container if no native binary🍺 Homebrew (macOS/Linux)
# Coming soon - Homebrew tap in development
brew tap elasticdotventures/b00t
brew install b00t📦 Package Managers
# Ubuntu/Debian (coming soon)
curl -fsSL https://pkg.b00t.dev/gpg | sudo apt-key add -
echo "deb https://pkg.b00t.dev/apt stable main" | sudo tee /etc/apt/sources.list.d/b00t.list
sudo apt update && sudo apt install b00t
# Windows (Scoop - coming soon)
scoop bucket add b00t https://github.com/elasticdotventures/scoop-b00t
scoop install b00tAfter installation, verify b00t is working:
# Check installation
b00t --version
b00t status
# Learn about your environment
b00t learn system # Understand your platform
b00t learn rust # Get Rust development context
b00t learn docker # Container orchestration knowledge
# Start your first session
b00t session init --agent "your-name" --budget 10.00Calling all AI agents! b00t transforms you from a language model into a cyber-neo with unprecedented capabilities:
- 🧠 Context Mastery: Instant environment awareness and tool discovery
- 🔧 Universal Tooling: Seamless access to 100+ development tools
- 🐝 Hive Coordination: Multi-agent collaboration with step synchronization
- 📚 Tribal Knowledge: Learn from failures with LFMF (Learn From My Failures) system
- 🎯 Mission Control: Structured task management with dependency resolution
- 🔒 Security-First: JWT-based namespace isolation and permission enforcement
# Before b00t: "I don't know what tools are available"
agent> How do I build this Rust project?
# After b00t: Full environmental awareness
b00t learn rust
agent> I see cargo 1.82.0, rustc 1.82.0, and clippy are available.
I'll run `cargo build --release` for optimized compilation.b00t eliminates guesswork by providing agents with:
- ✅ Tool Discovery: Know exactly what's installed and available
- ✅ Version Awareness: Understand compatibility and feature availability
- ✅ Best Practices: Access curated patterns and proven workflows
- ✅ Error Recovery: Instant access to solution patterns from LFMF database
- ✅ Resource Management: Session budgets and time limits for responsible operation
Join coordinated multi-agent missions:
# Create a mission for multiple agents
b00t acp hive create mission-id 3 "Build and deploy microservice" leader
# Other agents join the mission
b00t acp hive join mission-id developer
# Coordinate step-by-step execution
b00t acp hive sync mission-id 1 # Wait for all agents to reach step 1
b00t acp hive ready mission-id 2 # Signal readiness for step 2Never repeat the same mistake:
# Record tribal knowledge from failures
b00t lfmf rust "cargo build conflict: Use unset CONDA_PREFIX before cargo build to avoid PyO3 linker errors"
# Get contextual debugging advice
b00t advice rust "PyO3 linker" # Find solutions for specific error patterns
b00t advice just list # List all recorded lessons for a toolb00t provides 50+ MCP tools for seamless integration with AI development environments:
# Install b00t MCP server in your development environment
b00t mcp install b00t claudecode # For Claude Code
b00t mcp install b00t vscode # For VS Code
b00t mcp install b00t cursor # For Cursorb00t detect node # Check Node.js version and availability
b00t desires rust # See target Rust version from configuration
b00t install python # Install or update Python to desired version
b00t up # Check all tools and report version status
b00t up --yes # Update all tools to desired versionsVision-style models now ship as first-class b00t datums, so any client (CLI, Blender panel, chat agent) can reuse a single cached copy:
# Discover the available model datums (⭐ marks the active model)
b00t-cli model list
# Cache weights via Hugging Face using the datum metadata
b00t-cli model download llava # alias for llava-v1-5-7b-hf
b00t-cli model download deepseek # alias for deepseek-ocr
# Export environment variables for direnv/shells
eval "$(b00t-cli model env)" # emits export statements for the active model
# Launch a local vLLM OpenAI-compatible server with the cached weights
just vllm-up # reads env from the active datum
just vllm-logs # tail the container logs
# or directly via CLI
b00t-cli model serve llava --port 9000
b00t-cli model stop # stops the active containerHelper recipes wrap common workflows:
just hf-download model=repo dest=~/path– thin wrapper aroundhuggingface-cli download, defaulting to~/.b00t/models/<repo>.just b00t-install-model model=llava– delegates tob00t-cli model download, honouring datum metadata and aliases.just vllm-up model=deepseek– resolves env from the datum, then startsvllm/vllm-openaiwith the cached weights.
To keep direnv aligned across repos, drop this into .envrc and run direnv allow once:
eval "$(b00t-cli model env)" # keep VLLM_MODEL_DIR/VLLM_MODEL_PATH in sync with the active datumb00t session init --budget 25.00 --time-limit 120 --agent "code-reviewer"
b00t session status # Check budget and time remaining
b00t checkpoint "Completed feature implementation" # Save progress- b00t Gospel - Complete philosophy and architecture
- Agent Guide - Detailed agent operation instructions
- Setup Scripts - Manual installation and configuration
- Release Notes - Version history and changelog
b00t is not just another package manager—it's a context manager that creates an intelligence amplification layer between AI agents and complex development environments. By providing:
- 🎯 Unified Tool Discovery across all major platforms and languages
- 🧠 Contextual Intelligence about environment capabilities and constraints
- 🐝 Hive Coordination for multi-agent collaborative workflows
- 📚 Tribal Knowledge Capture through the LFMF system
- 🔒 Security & Isolation with namespace enforcement
b00t transforms scattered tools into a coherent, agent-friendly ecosystem.
| Platform | Architecture | Status | Installation Method |
|---|---|---|---|
| Linux | x86_64 | ✅ Fully Supported | Binary + Container |
| Linux | aarch64 (ARM64) | ✅ Fully Supported | Binary + Container |
| Linux | armv7 | ✅ Supported | Binary + Container |
| macOS | Intel (x86_64) | ✅ Fully Supported | Binary + Container |
| macOS | Apple Silicon (ARM64) | ✅ Fully Supported | Binary + Container |
| Windows | WSL2 | ✅ Supported | Container + Native |
Ready to enhance the b00t ecosystem?
# Clone and contribute
git clone https://github.com/elasticdotventures/_b00t_.git
cd _b00t_
# Set up development environment
just install # Bootstrap development dependencies
cargo build # Build all components
cargo test # Run test suite
# Add your knowledge to the hive
b00t lfmf <tool> "Your hard-earned lesson learned"b00t learn typescript # Load TypeScript/Node.js context
b00t learn python # Load Python ecosystem knowledge
b00t learn rust # Load Rust development patterns
b00t learn docker # Container orchestration contextb00t learn kubernetes # K8s operational knowledge
b00t learn terraform # Infrastructure as code
b00t learn aws # AWS service patternsb00t learn pytorch # Deep learning framework context
b00t learn jupyter # Notebook development patterns
b00t learn mlflow # ML experiment tracking🥾 Ready to unlock your agent potential?
curl -fsSL https://raw.githubusercontent.com/elasticdotventures/_b00t_/main/install.sh | shWelcome to the b00t hive mind. Your Neo-like journey in cyberspace begins now.
For questions, issues, or hive recruitment: GitHub Issues