The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through untrusted file inputs.
Ingestion points: The skill ingests data from a user-uploaded syllabus file via the read_file tool as defined in SKILL.md.
Boundary markers: There are no protective delimiters or instructions to ignore embedded commands when processing the syllabus content.
Capability inventory: The skill utilizes create_scheduled_task for SMS delivery and use_app_google_calendar for calendar management, providing a surface for misuse if malicious content is parsed from the file.
Sanitization: The protocol does not include any validation or sanitization steps to filter or escape potentially malicious text from the syllabus file before it is used to drive tool execution.

syllabus-reminders