The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses run_shell_command to execute system scheduling utilities such as schtasks on Windows and at or cron on Unix-based systems to manage user reminders.
[COMMAND_EXECUTION]: The skill attempts to execute a local Python script at scripts/calendar.py via run_shell_command to facilitate agenda synchronization.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to the processing of untrusted data from external sources.
Ingestion points: The agent reads and processes email content for categorization and drafting, and fetches external web content using google_web_search and web_fetch.
Boundary markers: There are no instructions providing clear delimiters or directives to ignore embedded commands in the processed external text.
Capability inventory: The skill has access to powerful tools including run_shell_command for system-level execution and save_memory for persistent state storage.
Sanitization: The skill instructions do not specify any validation, filtering, or sanitization of the data retrieved from external communications or web searches.

personal-assistant