The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing shell commands to interact with tmux and the system crontab.\n
scripts/tmux_run_job.py uses subprocess.run to execute commands inside a bash -lc shell within tmux windows.\n
scripts/tmux_healthcheck.py and scripts/tmux_gc.py use subprocess.run for tmux session and window management (list, kill, etc.).\n
scripts/tmux_healthcheck_cron.py executes crontab to read and modify the user's scheduled tasks.\n- [PERSISTENCE]: The skill provides automated maintenance through cron job installation.\n
scripts/tmux_healthcheck_cron.py contains logic to add, status, and remove an entry in the user's crontab (crontab -).\n
The cron entry executes a cleanup script (scripts/tmux_healthcheck.py) periodically (default every 15 minutes).\n- [DYNAMIC_EXECUTION]: The skill assembles and executes command strings at runtime.\n
scripts/tmux_run_job.py constructs a wrapped shell command including status tracking and tmux signaling logic, which is then passed to the shell.\n
scripts/tmux_healthcheck_cron.py dynamically builds the cron command string based on provided arguments like session names and log paths.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides mechanisms to ingest data from running jobs, creating a surface for indirect prompt injection.\n
Ingestion points: SKILL.md instructs the agent to use tmux capture-pane to read output from background jobs.\n
Boundary markers: The skill documentation does not specify the use of delimiters or instructions to ignore embedded commands in the captured output.\n
Capability inventory: The agent has capabilities to execute shell commands, manage tmux sessions, and modify crontab entries.\n
Sanitization: There is no evidence of sanitization or filtering applied to the captured terminal output before it is processed by the agent.

tmux