skills/b-open-io/prompts/Deck Creator/Gen Agent Trust Hub

Deck Creator

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The skill installs 'anthropics/theme-factory' using npx. Since 'anthropics' is an explicitly trusted organization in the security guidelines, this finding is downgraded to LOW severity.
  • COMMAND_EXECUTION (MEDIUM): The skill executes a local TypeScript script ('stitch-to-pdf.ts') via 'bun run' to create the final PDF. This involves filesystem access to read images and write documents.
  • PROMPT_INJECTION (MEDIUM): The skill has an indirect prompt injection surface. Evidence: 1. Ingestion points: 'Discovery' phase in 'SKILL.md' accepts untrusted documents/materials to inform content. 2. Boundary markers: Absent. 3. Capability inventory: Uses 'content-specialist' subagents for generation and 'bun' for file assembly. 4. Sanitization: Absent. External content could contain instructions that influence the subagent's planning logic.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 12:21 AM