skill-content-calendar-log
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing local Python scripts (
skills/shared/scripts/calendar_ops.py) to perform its core functions, including querying the calendar, adding events, and seeding holidays. This is standard functionality for this type of tool. - [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting and processing untrusted data.
- Ingestion points: Data enters the system via command-line arguments (
--title,--note,--url) and external files during theimport-eventscommand. - Boundary markers: The instructions do not specify any delimiters or sanitization steps when reading back the calendar data.
- Capability inventory: The skill has the ability to read and write to
outputs/_schedule.jsonand execute local scripts. - Sanitization: The skill instructions suggest that suggestions generated from the stored data should be relayed to the user "as is" (
原样转达用户), which could allow malicious content stored in the calendar (e.g., from an imported file) to influence the agent's behavior or deceive the user during subsequent "context" or "planning" phases.
Audit Metadata