skills/lovstudio/any2pdf/lov-any2pdf/Gen Agent Trust Hub

lov-any2pdf

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches emoji assets from Cloudflare's well-known CDN (cdnjs.cloudflare.com) to support emoji rendering.
  • [EXTERNAL_DOWNLOADS]: Downloads remote images specified in the input Markdown documents via urllib.request.urlopen.
  • [DATA_EXFILTRATION]: Resolves and reads local file paths and file:// URLs provided in the Markdown source to embed images in the generated PDF, which constitutes local file system access.
  • [PROMPT_INJECTION]: The skill processes untrusted Markdown content, creating an indirect prompt injection surface, particularly when the agent is instructed to read the document for AI-assisted cover generation.
  • Ingestion points: Markdown source file provided as input for the conversion process.
  • Boundary markers: None; the skill parses the document structure directly without delimiters to ignore embedded instructions.
  • Capability inventory: Local file system read/write access, network read capabilities, and the triggering of external image generation tools.
  • Sanitization: Employs standard URL and path parsing but lacks content-based filtering for malicious instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 12:20 PM
Security Audit — agent-trust-hub — lov-any2pdf