Ailoop is a command-line tool that enables AI agents to communicate with human users through structured interactions. It supports both direct mode (for single-agent scenarios) and server mode (for multi-agent environments), providing a seamless bridge between AI automation and human oversight.

• Ask questions and wait for human responses
• Request authorization for critical actions
• Send notifications with priority levels
• Display images and navigate to URLs
• Channel-based isolation for multi-agent workflows

• Centralized server for multi-agent environments
• Interactive terminal UI with real-time monitoring
• WebSocket-based communication
• Channel management and message history
• Web interface for remote monitoring

• Stream agent output to centralized server
• Support for multiple agent types (Cursor, JSONL, etc.)
• Real-time message broadcasting
• Message history and channel switching

Download the latest release from GitHub Releases:

• Linux GNU: ailoop-x86_64-unknown-linux-gnu.tar.gz
• Linux MUSL: ailoop-x86_64-unknown-linux-musl.tar.gz
• Windows: ailoop-x86_64-pc-windows-msvc.zip

Extract and add to your PATH, then make executable on Unix systems.

brew install goailoop/cli/ailoop

scoop bucket add goailoop https://github.com/goailoop/scoop
scoop install ailoop

ailoop --version

ailoop ask "What is the best approach for this task?"
ailoop ask "Should we proceed?" --timeout 60
ailoop ask "Review this code change" --channel dev-review

ailoop authorize "Deploy version 1.2.3 to production"
ailoop authorize "Delete user data" --timeout 300 --channel admin-ops

Note: If no response is received within the timeout period, authorization defaults to DENIED for security.

ailoop say "Build completed successfully"
ailoop say "System alert: High CPU usage detected" --priority high
ailoop say "Critical error occurred" --priority urgent --channel alerts

# Stream agent output to server
ailoop forward --channel my-agent

# Forward from file
ailoop forward --input agent-output.jsonl --channel my-agent

Ask a question and wait for a human response.

ailoop ask "What is the best approach?"              # Simple question
ailoop ask "Should we proceed?" --timeout 60         # With timeout
ailoop ask "Review this code" --channel dev-review  # Specific channel
ailoop ask "What is your name?" --json              # JSON output

Options: --channel, -c, --timeout, -t, --server, --json

Request human approval for critical actions.

ailoop authorize "Deploy to production"
ailoop authorize "Delete user data" --timeout 300 --channel admin-ops
ailoop authorize "Execute operation" --json

Options: --channel, -c, --timeout, -t, --server, --json

Send notification messages to human users.

ailoop say "Build completed successfully"
ailoop say "High CPU usage" --priority high
ailoop say "Critical error" --priority urgent --channel alerts

Options: --channel, -c, --priority, --server

Available priorities: low, normal, high, urgent

Run ailoop in server mode for multi-agent environments.

ailoop serve                    # Default port 8080
ailoop serve --port 9000        # Custom port
ailoop serve --host 0.0.0.0 --port 8080 --channel default

The server provides an interactive terminal UI showing:

• Server status and metrics
• Queue size and active connections
• Recent activity

Press Ctrl+C to stop the server.

Stream agent output to centralized server.

ailoop forward --channel my-agent
ailoop forward --input agent-output.jsonl --channel my-agent
ailoop forward --transport websocket --url ws://localhost:8080

Options: --channel, -c, --agent-type, --format, --transport, --url, --input, --output, --client-id

Set up your configuration interactively.

ailoop config --init
ailoop config --init --config-file ~/.config/ailoop/custom.toml

The configuration wizard will prompt you for:

• Default timeout for questions
• Default channel name
• Log level (error/warn/info/debug/trace)
• Server bind address and port

Configuration is saved in TOML format at ~/.config/ailoop/config.toml by default.

# Agent asks for code review
ailoop ask "Please review PR #123" --channel code-review --timeout 3600

# Agent requests approval
ailoop authorize "Merge PR #123 to main branch" --channel code-review

# Start server for deployment team
ailoop serve --port 8080 --channel deployments

# In another terminal, request deployment approval
ailoop authorize "Deploy v2.1.0 to production" \
  --server http://localhost:8080 \
  --channel deployments \
  --timeout 600

# Send urgent alert
ailoop say "Database connection pool exhausted!" \
  --priority urgent \
  --channel monitoring

# Show monitoring dashboard
ailoop navigate https://monitoring.example.com/dashboard \
  --channel monitoring

# Start server
ailoop serve --port 8080

# Forward agent output from stdin
agent-command | ailoop forward --channel my-agent

# Forward from file
ailoop forward --input agent-output.jsonl --channel my-agent

Channels provide isolation between different workflows or agent instances. Each channel maintains its own message queue and connections.

• Channel names must be 1-64 characters
• Use lowercase alphanumeric characters, hyphens, or underscores
• Must start with a letter or digit
• Default channel is public

Examples of valid channel names:

• production
• dev-review
• analytics_team
• channel-123

• Ensure binary is in your PATH
• Make executable on Unix: chmod +x ailoop

• Check if the port is already in use: lsof -i :8080 (Linux) or netstat -an | grep 8080 (Windows)
• Ensure you have permission to bind to the specified port (ports < 1024 require root/admin)
• Check firewall settings

• Increase timeout value if operations take longer than expected
• Use --timeout 0 to wait indefinitely (not recommended for production)

• Ensure channel names follow the naming convention
• Channel names are case-sensitive
• Avoid special characters except hyphens and underscores

• Verify server is running: ailoop serve
• Check server URL format: http://localhost:8080 or ws://localhost:8080 for WebSocket
• Ensure firewall allows connections on the specified port

• Version: ailoop --version
• Command help: ailoop <command> --help
• Documentation: GitHub Repository

Apache License, Version 2.0 - See LICENSE file for details.

Need help? Open an issue on GitHub.