The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary function is to facilitate the execution of arbitrary shell commands. It acts as a bridge between the agent and a background daemon, allowing the agent to launch scripts that continue beyond a single conversation iteration.
[PERSISTENCE]: The schedule command allows the agent to establish recurring execution of commands at fixed intervals. While this is the intended purpose of the skill, it represents a persistence mechanism within the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to read and process script.result messages and log files generated by executed scripts. This creates a surface where output from untrusted or compromised external scripts could potentially contain instructions aimed at influencing the agent's future actions.
Ingestion points: script.result messages and log files (referenced via log_path in results) processed by the agent.
Boundary markers: Not explicitly defined in instructions; the agent is instructed to "Process the result" to determine if a script should be resumed or removed.
Capability inventory: The skill can execute, schedule, rerun, and remove shell scripts via a Unix socket.
Sanitization: No explicit sanitization or filtering of script output is mentioned in the skill instructions.

scripts