roadmap-timeline
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests data from SharePoint and user input to generate HTML reports, representing a surface for indirect prompt injection.\n
- Ingestion points: SharePoint lists via tool calls and user-provided text or table data (SKILL.md).\n
- Boundary markers: The skill requires data validation and the construction of a structured JSON document prior to rendering (Step 4).\n
- Capability inventory: The skill generates output files containing HTML and inline JavaScript (Step 5).\n
- Sanitization: While the skill enforces a validation checklist for data integrity, it does not explicitly mandate HTML entity encoding for the user-provided text fields embedded in the final report.
Audit Metadata