education-cloud-academic-calendar-generate

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and parse untrusted data from external sources, including natural language prompts and uploaded documents (PDF, CSV, and XLSX) to extract academic calendar dates.
  • Ingestion points: Workflow Step 2 identifies that the agent extracts calendar dates from user-provided documents or natural language descriptions.
  • Boundary markers: There are no explicit instructions or delimiters provided to the agent to isolate the untrusted document content from the internal instructions during the parsing phase.
  • Capability inventory: The skill utilizes Salesforce REST API capabilities (POST, GET, PATCH) to create and modify records in the target organization.
  • Sanitization: The skill mitigates risks by implementing a deterministic date validation script (scripts/validate_calendar_dates.py) and a mandatory human-in-the-loop confirmation step (Workflow Step 6) where the proposed calendar structure must be reviewed and approved by the user before any records are created.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 07:35 AM
Security Audit — agent-trust-hub — education-cloud-academic-calendar-generate