fsi-strip-profile

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions for the agent to use soffice (LibreOffice) and pdftoppm (Poppler) commands. These tools are used to convert generated PowerPoint files into images, enabling a mandatory visual review step to ensure layout quality and prevent text collisions.
  • [PROMPT_INJECTION]: The skill processes untrusted external data from financial sources, creating a potential indirect prompt injection surface.
  • Ingestion points: Company filings, investor presentations, and financial news reports (specified in the Research section of SKILL.md).
  • Boundary markers: A mandatory user-alignment step requires the agent to print an outline for review before slide generation begins, serving as a manual checkpoint.
  • Capability inventory: Shell access for image conversion tools and PptxGenJS for presentation generation.
  • Sanitization: External data is filtered and normalized into predefined financial metrics and bullet points, which inherently limits the influence of unstructured prose.
  • [DATA_EXFILTRATION]: Research workflows target established financial databases and regulatory platforms (e.g., SEC EDGAR, Bloomberg, FactSet). These activities are restricted to gathering market metrics for profile creation and do not involve exfiltration of sensitive system credentials or local files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 01:28 PM
Security Audit — agent-trust-hub — fsi-strip-profile