Bytebot Documentation

The Bytebot REST API is categorized into several functional areas:

• Computer Use: Execute desktop automation actions such as mouse movements, clicks, keyboard input, and screenshots.
• Input Tracking: Control and stream keyboard and mouse events from the desktop.
• Usage Examples: Access code examples and snippets for common automation scenarios.

Bytebot differs from traditional RPA by using visual intelligence and natural language instead of brittle element mapping and complex scripting.

To initiate a new task for the agent:

1. Enter a description for the task in the Task Management Panel.
2. Click the "Start Task" button or press Enter.

The default configuration prioritizes ease of use. For production environments, implement these security hardening steps:

1. Change Default Passwords: Use passwd bytebot to set a new user password.
2. Limit Network Access: Use environment variables to whitelist domains or restrict port access to specific IP ranges (e.g., 10.0.0.0/8).

For enterprise deployments, you can install Bytebot on Kubernetes using Helm. You must provide an AI provider API key via the --set flag.

# Clone the repository
git clone https://github.com/bytebot-ai/bytebot.git
cd bytebot

# Install with Helm
helm install bytebot ./helm \
  --set agent.env.ANTHROPIC_API_KEY=sk-ant-...

Bytebot can automatically fill passwords and handle 2FA (TOTP) if you install a password manager extension via the Desktop view. 1Password is the recommended option.

Setup Steps:

1. Install Extension: In the Bytebot UI, go to the Desktop tab, open Firefox, and install the 1Password extension from the Firefox Add-ons store. Sign in to your account.
2. Configure Vaults: In your 1Password admin panel, create a vault (e.g., Bytebot Automation), add the necessary credentials, and share it with Bytebot's account using read-only permissions.
3. Enable Auto-fill: The extension will then automatically detect login forms, fill credentials, handle 2FA codes, and submit forms during tasks.

The Desktop tab on the main page provides free-ranging access to the desktop environment outside of specific tasks.

Key characteristics:

• No Recording: Unlike Takeover Mode, nothing is recorded in this mode.
• Use Cases: Ideal for installing programs, logging into apps/websites, or general desktop maintenance and setup.

You can self-host the full Bytebot stack using Docker Compose.

Prerequisites:

• Docker ≥ 20.10
• Docker Compose
• 4GB+ RAM
• An API key from Anthropic, OpenAI, or Google.

Setup Steps:

1. Clone the repository and enter the directory:

   git clone https://github.com/bytebot-ai/bytebot.git
   cd bytebot

2. Configure your AI provider by creating a .env file in the docker/ directory (choose one):

   echo "ANTHROPIC_API_KEY=your_api_key_here" > docker/.env
   # OR
   echo "OPENAI_API_KEY=your_api_key_here" > docker/.env
   # OR
   echo "GEMINI_API_KEY=your_api_key_here" > docker/.env

3. Start the stack:

   docker-compose -f docker/docker-compose.yml up -d

4. Access the UI at http://localhost:9992.

You can manually interrupt Bytebot during task execution to guide it, resolve issues, or correct its path.

Steps to manually take control:

1. Open Task Detail View: While Bytebot is working, click on the specific task to open its detail view.
2. Click Takeover Button: Hit the takeover button to interrupt the agent and take control of the desktop.
3. Guide Bytebot: Perform the necessary actions (clicks, drags, typing, etc.) to resolve the obstacle.
4. Release Control: Click to release control, allowing Bytebot to resume the task from your current state.