The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to interact with the host system's tmux utility. It provides examples for creating sessions, listing panes, and sending keys. The included helper scripts, scripts/find-sessions.sh and scripts/wait-for-text.sh, use the tmux CLI to perform polling and discovery operations.
[PROMPT_INJECTION]: The skill lacks safeguards against indirect prompt injection, as it encourages the agent to capture and parse terminal output that may originate from untrusted external sources (e.g., website content, logs, or interactive CLI outputs).
Ingestion points: tmux capture-pane is utilized in SKILL.md and scripts/wait-for-text.sh to retrieve the contents of terminal windows.
Boundary markers: There are no delimiters or specific instructions provided to the agent to differentiate between system status messages and untrusted data within the tmux scrollback.
Capability inventory: The agent has the capability to execute commands in sub-shells via tmux send-keys and manage background processes.
Sanitization: The skill does not perform any validation or escaping of the data captured from the terminal before the agent processes it.

tmux