The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides templates to drive interactive CLIs using tmux and Python's pty module. These operations are performed locally for testing and debugging purposes. No patterns for arbitrary command injection or unauthorized remote execution were found.
[SAFE]: The skill specifies guardrails that prevent the use of credentials, destructive commands, or hard-coded paths from other repositories. It also instructs the user to clean up temporary sessions and files.
[PROMPT_INJECTION]: A potential surface for indirect prompt injection exists because the skill reads and acts upon terminal output to synchronize input. This represents a functional surface documented for awareness but is assessed as safe given the local testing context and manual harness creation.\n
Ingestion points: Terminal output captured via tmux capture-pane or Python os.read as described in SKILL.md.\n
Boundary markers: Absent.\n
Capability inventory: tmux send-keys, subprocess.Popen, and os.write (interaction with local processes).\n
Sanitization: Absent.

control-cli