Run commands
Use alpacon exec for commands that can finish without terminal input.
Choose a command workflow
To check a service, collect output, or run an automation script, start with Run a command. For a script whose contents need review, use verified file execution.
Commands or the web terminal
| What you need | alpacon exec | Websh |
|---|---|---|
| Run a command and collect its result | Non-interactive execution; no stdin | An interactive terminal |
| Answer a prompt or use a terminal application | Use Websh instead | Supports terminal input |
| Work-session feature | Execute commands (command) | Web terminal (websh) |
| AI agent access | An agent work session can run commands | An agent work session cannot include Websh |
| Token authentication | Use a service token for automation | Tokens cannot open an interactive terminal |
See Websh for interactive work.
What a command needs
With browser login, select an active work session that includes the target server and the command feature (command). Run as an account the session permits. An agent work session runs only as its approved account.
A service token can run non-interactive commands without a work session, subject to its scopes and command ACLs. New personal API tokens cannot hold command-execution or file-transfer scopes. Existing tokens with those scopes keep working without a work session. Use a service token for automation and CI.
Where results appear
Open Command history to see the requester, server, account, result, and Rejection reason. The work session’s History tab also shows command entries; expand an entry to read its result. From the CLI, use alpacon exec ls to find a command and alpacon exec logs JOB_ID to fetch its result.
Live activity shows the last 16,384 characters of output while a command runs. The stored result keeps up to 1 MiB; longer output loses its middle, retaining the beginning and end with a truncation marker.
Limits
| Limit | Value |
|---|---|
| Command line | The command line can contain up to 2048 characters. Shorten the command or use verified file execution for a script. |
Purpose (--purpose) | Up to 2000 characters. A longer purpose is refused. Describe the reason for this command within the limit. |
Script contents (--file) | A non-empty UTF-8 script of at most 65,536 bytes (64 KB). Split a larger script or have it read data from a separate file. |
| File execution command line | The interpreter, script path, and arguments together must fit within 2048 characters. Shorten paths or arguments; have the script read bulky input from a file. |
| Live output | Live activity shows the last 16,384 characters. Open the stored command result for more output. |
| Stored result | Up to 1 MiB. Longer output is cut in the middle, keeping the beginning and end. Have the command write a file on the server if you need all of a large result. |
| Run time | A command runs on the server for up to 30 minutes. Split work that cannot finish within 30 minutes. |
| Foreground wait | After 10 minutes without a result following acknowledgment or delivery, foreground exec reports the command as stuck and exits 1. The command keeps running on the server for up to 30 minutes. Use --detach and alpacon exec logs JOB_ID for work that may take longer than about 10 minutes. |
CLI status wait (ALPACON_EXEC_TIMEOUT) | How long the CLI keeps waiting when it cannot get a running status; default 30m. Each in-progress status resets this wait. It is not a run-time limit. Check the job before retrying: giving up waiting does not cancel it. |
Approval wait (--wait-approval) | The wait for approval defaults to 5 minutes; accepts a duration such as 30m. Use --wait or set --wait-approval when you want the CLI to wait for approval. |