Build, deploy, and manage autonomous AI agents with built-in memory management, safe tool sandboxing, and multi-provider LLM support.
Everything you need to build and deploy production-ready AI agents
Command filtering blocks dangerous operations. Input validation prevents injection attacks. Optional Docker isolation for safe execution.
Real-time conversations with any LLM provider. Rich terminal UI with markdown rendering, syntax highlighting, and chat history.
Switch between OpenAI, Claude, Gemini, Cohere, GitHub Models, and local Ollama models with a single flag.
SQLite-backed agent registry with status tracking, log aggregation, and graceful shutdown with SIGTERM/SIGKILL support.
Configurable timeouts prevent runaway processes. Resource limits ensure agents don't consume excessive system resources.
Define agents with simple YAML configuration files. Specify models, security settings, and custom blocked commands.
Connect to your favorite AI models with zero configuration
Simple installation process for Linux and Windows
git clone https://github.com/agents-os/agentos.git
cd agentos
python3 install_linux.py
cp .env.example .env && nano .env
git clone https://github.com/agents-os/agentos.git
cd agentos
python install_windows.py
copy .env.example .env && notepad .env
Get your first agent running in under 5 minutes
name: my_assistant
model_provider: github
model_version: openai/gpt-4o-mini
isolated: false
DESTRUCTIVE_COMMANDS:
- rm
- sudo
- rmdirSave as agent.yaml
# Run a task
agentos run agent.yaml --task "create hello world"
# Monitor agents
agentos ps
# View logs
agentos logs my_assistant# Chat with Claude
agentos chat --provider claude
# Use local Ollama (no API key!)
agentos chat --provider ollama
# Custom settings
agentos chat --provider openai --model gpt-4Powerful command-line interface for complete control
agentos run <manifest> --task "<task>"
Run an agent with a specific task
agentos chat [--provider <name>]
Start interactive chat mode
agentos ps
List all running agents
agentos logs <agent_name> [--tail 50]
View agent execution logs
agentos stop <agent_name>
Stop a running agent
agentos prune
Remove stopped agents
Join the community building the future of AI agents