Exec

The exec capability runs arbitrary commands on the target and returns their output as a task result. It is available for all transports and agent modes (beacon and session).

Usage

sentrava [session] > shell <command>

Despite using the shell CLI keyword, exec dispatches to the agent's exec capability which runs the command directly (not through a shell). To use shell features (pipes, redirects, glob expansion), invoke the shell explicitly:

sentrava [session] > shell bash -c "cat /etc/passwd | grep root"
sentrava [session] > shell cmd.exe /c "dir C:\Users"

Examples

sentrava [session] > shell id
sentrava [session] > shell uname -a
sentrava [session] > shell hostname
sentrava [session] > shell cat /etc/os-release
sentrava [session] > shell bash -c "find /home -name '*.ssh' -type d 2>/dev/null"

Windows:

sentrava [session] > shell whoami
sentrava [session] > shell ipconfig /all
sentrava [session] > shell net user
sentrava [session] > shell cmd.exe /c "systeminfo"

Output

Command output (stdout and stderr combined) is captured and stored as a task result in the server database. Results survive agent restart and are available for review after the session ends.

Timeout

Commands inherit the agent's task timeout. Long-running commands that exceed the timeout are killed by the agent.

Task persistence

Exec tasks are persisted to the database. View past results:

sentrava [session] > tasks list

Difference from interactive shell

The shell command is a one-shot exec: it runs the command, collects output, and returns. It does not provide a persistent shell environment. Environment changes (like cd) do not persist between invocations.

For an interactive shell with persistent state, use shell-interactive (mTLS session mode, Linux agents only).