Skip to main content

​
Getting Started

Get FERAL running locally in under five minutes. By the end of this page you will have a brain server, a web dashboard, and your first conversation.

​
Prerequisites

  • Python 3.11+ — python3 --version
  • An LLM API key (OpenAI, Anthropic, Gemini, Groq) or a local Ollama instance

​
Install

pip install feral-ai[llm]
This installs the feral CLI, the FastAPI brain server, and the bundled web UI.

​
Setup

Run the interactive setup wizard: 
feral setup
The wizard walks you through:
StepWhat It Configures
LLM ProviderChoose OpenAI, Anthropic, Gemini, Groq, or Ollama (free/local). API key is validated live.
Agent IdentityName, personality, voice settings, and behavioral rules.
Skills & ToolsEnable computer use, web search, vision, hardware control. Add keys for Tavily, Spotify, etc.
FeaturesVoice mode (realtime / whisper / disabled), streaming, proactive behavior, wake word.
All configuration is written to ~/.feral/. No cloud account needed.

​
Start

feral start
This launches the brain and serves the web dashboard at http://localhost:9090. Open the URL in your browser. Type a message or click the microphone for voice.

​
First Chat (CLI)

You can also talk to FERAL from the terminal: 
feral "What files are in my home directory?"
feral "Search the web for latest AI news"
feral "Remember that my favorite color is blue"
feral "What's my favorite color?"

​
First Chat (SDK)

from feral_sdk import FeralClient

async with FeralClient("http://localhost:9090") as client:
    reply = await client.chat("Hello! What can you do?")
    print(reply)

​
Docker Alternative

If you prefer Docker: 
git clone https://github.com/FERAL-AI/FERAL-AI.git && cd FERAL-AI
cp .env.example .env   # fill in your API keys
docker compose up -d
ServiceURL
Brain + APIhttp://localhost:9090
Web UIhttp://localhost:3000
Skill Registryhttp://localhost:8080

​
What’s Next