connect-ai-googlecalendar

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is entirely informational, providing SQL examples and metadata for a specific database driver. No malicious scripts, unauthorized network operations, or persistence mechanisms were detected.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external event data (Summaries, Descriptions, and Locations) from Google Calendar. This creates a standard ingestion surface where data controlled by other users could theoretically contain instructions intended to influence the agent.
  • Ingestion points: SQL query results from the AllCalendars and Calendars tables (SKILL.md).
  • Boundary markers: The provided SQL query templates and descriptions do not include specific delimiters or 'ignore' instructions for the agent when processing retrieved text fields.
  • Capability inventory: The agent has data modification capabilities through SQL UPDATE statements and stored procedures such as ImportEvent and QuickAddEvent (SKILL.md).
  • Sanitization: No explicit sanitization or content validation steps are defined within the skill instructions for handling retrieved calendar data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 02:09 PM
Security Audit — agent-trust-hub — connect-ai-googlecalendar