Noxgild
Using Noxgild

Terminal and Processes

Run bounded terminal commands and manage persistent processes through Noxgild.

The Machine layer supports both bounded shell commands and managed long-running processes.

Bounded commands

A bounded shell command can run inside an authorized workspace with limits on runtime and returned output.

Use this for commands expected to finish, such as building a project, running tests, inspecting a tool version, or generating a known result.

Managed processes

For work that stays alive, Noxgild can start a managed process and then inspect, feed input to, wait for, or terminate that process.

Process start supports a shell command string or a direct executable plus argument vector. Direct executable mode does not route the process through a command shell.

Process identity

Status and listing results can include the managed process ID, execution mode, executable, arguments, and completion state so the client can target the right process.

Termination

Process termination is destructive and requires exact per-command approval before execution.

Retry safety

Process actions do not all share the same retry behavior. Noxgild takes the idempotency and verification rules from the capability catalog and treats uncertain side effects cautiously.

On this page