decision-toolkit

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user inputs or extracted meeting transcripts and interpolates it into generated HTML and Markdown templates using placeholders like {{CONTEXT}} and {{OPTIONS}}. The instructions do not define boundary markers or explicit prompts to ignore embedded instructions within that ingested data.
  • Ingestion points: User-provided context and document/transcript extraction defined in Step 1 of SKILL.md.
  • Boundary markers: Absent in the interpolation templates.
  • Capability inventory: Generation of interactive HTML/JS tools and formatted Markdown decision records.
  • Sanitization: No instructions provided to sanitize or escape ingested content before template rendering.
  • [EXTERNAL_DOWNLOADS]: The interactive decision guide template (templates/decision-guide-template.html) fetches typography resources from Google's official web font service (fonts.googleapis.com). This is a well-known service used for interface styling and does not involve executable code downloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 04:42 AM