The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands on remote hosts using tmux send-keys. It includes established patterns for handling sudo prompts, interactive CLI tools, and sequential command execution, all within the context of managing active remote panes.
[DATA_EXFILTRATION]: Provides instructions for capturing remote terminal output via tmux capture-pane and transferring files using scp and rsync. These features facilitate moving data from remote hosts to the local machine for processing, which is consistent with the skill's role as a management tool.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests data from external terminal output. 1. Ingestion points: Terminal output is read from remote panes using tmux capture-pane in several workflow examples. 2. Boundary markers: The skill utilizes unique marker strings (UUID/timestamp based) to detect command completion, providing a synchronization boundary but no content-based isolation. 3. Capability inventory: The skill can execute arbitrary commands, manage terminal sessions, and perform file transfers. 4. Sanitization: No mechanisms for sanitizing or escaping the content of the captured output are described before the data is processed by the agent.

Remote SSH via tmux