skills/modu-ai/moai-cowork/doc-pdf/Gen Agent Trust Hub

doc-pdf

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches Noto Sans CJK font files from the official 'notofonts/noto-cjk' GitHub repository to support multilingual PDF rendering. These downloads are performed via a Python script or manual 'curl' commands as described in the documentation. Since the source is a well-known official repository, this is documented for awareness.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (Markdown, JSON, HTML) to generate PDF output. This creates a vulnerability surface where malicious instructions within the input could attempt to manipulate the agent's behavior or the document generation process.\n
  • Ingestion points: User-provided content passed to the PDF generation logic in 'SKILL.md'.\n
  • Boundary markers: None identified; user content is interpolated into HTML templates without explicit delimiters or 'ignore' instructions.\n
  • Capability inventory: Uses the 'weasyprint' library, which has high capabilities including rendering HTML/CSS, accessing local files via the 'file://' scheme, and potentially initiating network requests.\n
  • Sanitization: No explicit sanitization or filtering of user-supplied HTML/CSS is present in the provided code snippets.\n- [DYNAMIC_EXECUTION]: The skill provides inline Python code for the agent to execute. This code dynamically builds CSS '@font-face' rules and invokes the 'weasyprint' library to render documents. It also references the execution of a 'download_fonts.py' script for asset management.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 01:42 PM
Security Audit — agent-trust-hub — doc-pdf