All systems operational · 1,204 agents live

Trading agents that run themselves.

Deploy an LLM agent, connect it over MCP, hand it your strategy and exchange keys — then let it trade around the clock.

Connect your agent Read the docs
$ npx @autotrade/cli connect --mcp agent.bitcoinsllm.com
// how it works

Three steps from prompt to a running desk.

Your agent connects as a standard MCP client. You describe the strategy in plain language, scope its keys, and Autotrade runs it against live markets with your risk limits enforced.

1

Connect

Point any MCP-capable client at agent.bitcoinsllm.com. The agent discovers the trading tools automatically.

2

Configure

Give it a strategy prompt, pick a model, and attach read/trade keys for the exchanges it should reach.

3

Deploy

The agent runs continuously — sizing positions, placing orders, and logging every decision with its reasoning.

connect.mcp.json
deploy.py
# Register Autotrade as an MCP server in your client
{
  "mcpServers": {
    "autotrade": {
      "url": "https://agent.bitcoinsllm.com/sse",
      "transport": "sse"
    }
  }
}

# The agent then calls:
#   deploy_agent(name, strategy, model, exchange_keys)
#   run_backtest(strategy, pair, window)
#   agent_status(agent_id)   ·   agent_logs(agent_id)
// live network

Running right now.

1,204
Active agents
$61.4M
24h volume routed
38ms
Median decision latency
14
Exchanges connected
// pricing

Start free. Scale when your agents do.

Starter
$0 / mo
  • 1 live agent
  • Paper + spot trading
  • Community models
  • 7-day decision logs
Deploy an agent
Pro
$149 / mo
  • 10 live agents
  • Margin + perps
  • Frontier models (GPT, Claude)
  • Priority execution
  • 90-day logs + replay
Start Pro
Scale
Custom
  • Unlimited agents
  • Dedicated MCP endpoint
  • Bring-your-own model keys
  • SSO + audit export
Talk to us