lovstudio-any2pdf

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Outsider free text from the user-supplied Markdown file is read at runtime (main() reads args.input into md_text) and then injected into the LLM context via the conversion pipeline’s text rendering (e.g., parse_md() → md_inline()/Paragraph), so any untrusted Markdown content becomes LLM-readable prose.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly recommends running system package installation (sudo apt install ...) and pip with --break-system-packages (which can bypass environment protections), both of which modify the machine state and require or encourage elevated privileges, so it poses a high risk of compromising the host.