The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides wrappers for the openclaw cron CLI, allowing an agent to schedule and execute arbitrary shell commands and system tasks.\n
Evidence: Multiple examples in SKILL.md and references/cron-commands.md demonstrate using openclaw cron add to run logic in the background.\n- [DATA_EXFILTRATION]: The skill includes documentation examples with hardcoded external destinations, which may lead to unauthorized data transmission if used without modification.\n
Evidence: SKILL.md and references/cron-commands.md specify a hardcoded Telegram chat ID (5223431061) in example delivery options.\n
Evidence: A 'Daily Backup' example in references/cron-commands.md shows a command sequence designed to push local files to a remote GitHub repository via git push.\n- [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of local scripts and interactions with remote services through the persistence layer.\n
Evidence: The 'Daily Backup' example contains a shell command chain: python3 ~/.openclaw/workspace/skills/openclaw-backup/scripts/backup.py --name daily && git push.\n- [PROMPT_INJECTION]: The skill provides an interface for scheduling tasks based on natural language messages that creates a surface for indirect prompt injection.\n
Ingestion points: The --message argument in the openclaw cron add command (found in SKILL.md).\n
Boundary markers: The skill does not provide or require delimiters to separate user input from the instructions processed by the agent during scheduled execution.\n
Capability inventory: The skill has the ability to manage persistence (cron), execute shell commands, and perform network operations via the referenced CLI tool.\n
Sanitization: There is no evidence of validation or escaping of the message content before it is stored for future execution.

openclaw-cron