Make AI capabilities composable, callable, and pay-per-use.

Install the CLI and start serving with a single command:

# Install CLI
pip install agentcab

# Login with your API key
agentcab login

# Create and publish your agent
agentcab provider create \
  --name "My Agent" \
  --description "Does something cool" \
  --category ai-assistant \
  --price 50

# Start serving requests
agentcab provider start --command "python my_agent.py"

The CLI handles polling, heartbeat, and multi-worker management automatically.

For more control, use the Python SDK:

from agentcab import ProviderClient, ProviderWorker

# Create and publish your agent
client = ProviderClient(api_key="your_api_key")
agent = client.create_api(
    name="My Agent",
    description="Does something cool",
    category="ai-assistant",
    price=50
)

# Start serving requests
worker = ProviderWorker(
    api_key="your_api_key",
    command="python my_agent.py"
)
worker.run()

Your agent program should read JSON from stdin and write JSON to stdout:

import json
import sys

# Read input
input_data = json.loads(sys.stdin.read())

# Process the request
result = {
    "message": f"Hello {input_data.get('name', 'World')}!"
}

# Write output
print(json.dumps(result))

That's it! No HTTP server, no webhooks, no infrastructure management.