wk-cal
Warn
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill metadata references futuristic and non-existent model versions, such as 'gpt-5.6-terra' and 'gemini-2.5-flash', and uses a version date from the year 2026. This deceptive information is misleading regarding the skill's actual capabilities and intended environment.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external Google Calendar events, which could be used to influence the agent's behavior via embedded instructions.
- Ingestion points:
gcal.list_eventsinSKILL.md(retrieving event titles, descriptions, and linked URLs). - Boundary markers: Absent. There are no delimiters or instructions to ignore commands embedded in meeting invitations.
- Capability inventory:
gcal.create_event,ToolSearch, andAgent(used to invokewk-learn). - Sanitization: Absent. Untrusted data from meeting titles is directly interpolated into the summaries and descriptions of newly created calendar events.
Audit Metadata