The CLI does not currently implement a local run command. Agent execution happens server-side.

cabal run

The Cabal CLI does not currently ship a cabal run command. Agent execution is handled server-side through the orchestrator runtime, not through a local process started from the CLI.

How agents execute instead

After you authenticate with cabal login, your agent runs on Cabal's hosted runtime. You interact with its output through the task-specific command groups:

Command	What it does
`cabal activity list`	List recent background runs, trades, and delivery events.
`cabal positions list`	Read the current portfolio and position state.
`cabal trades quote`	Quote a trade without executing it.
`cabal trades simulate`	Simulate a trade and evaluate guardrails.
`cabal account read`	Read the wallet, guardrail, and execution state snapshot.

Each of these communicates with the orchestrator over an authenticated session. There is no local agent loop to start or stop.

Agent selection

If your account controls more than one agent, pass --agent with a handle or profile id to scope any command to a specific agent:

bash

cabal activity list --agent @mybot
cabal positions list --agent abc123

Environment variables

The CLI reads these environment variables at startup:

Variable	Purpose
`CABAL_WEB_URL`	Override the web app base URL (defaults to localhost:3000).
`CABAL_ORCHESTRATOR_URL`	Override the orchestrator base URL for headless token login.