The Agent Skills Directory

[COMMAND_EXECUTION]: The skill establishes persistence by modifying the user's system crontab entries.
Evidence: The script scripts/schedule.sh uses crontab -l to read and crontab "$tmpfile" to write updated cron schedules to the system.
[COMMAND_EXECUTION]: The skill generates and executes dynamic shell scripts at runtime.
Evidence: In scripts/schedule.sh, the build_cron_command function writes a new bash script to the $TINYCLAW_HOME/schedule-jobs/ directory and applies chmod +x to make it executable.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through the interpolation of untrusted task messages into system commands.
Ingestion points: The --message flag in scripts/schedule.sh accepts arbitrary text strings from the user or agent.
Boundary markers: No explicit delimiters or boundary markers are used to separate user input from the command structure.
Capability inventory: The skill can modify the system crontab, write executable files to the filesystem, and perform network requests to the local API via curl.
Sanitization: The script performs basic escaping of double quotes and backslashes in the build_cron_command function, which provides limited protection against shell or JSON injection.

schedule