huggingface-paper-publisher

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill manages authentication securely by utilizing the HF_TOKEN environment variable, avoiding hardcoded secrets and adhering to standard API practices.
  • [SAFE]: The Python script implements robust input validation for arXiv identifiers using strict regular expression patterns (_clean_arxiv_id), which prevents common attack vectors like path traversal or command injection via malformed IDs.
  • [SAFE]: Data fetched from external sources (such as the arXiv API) and user-supplied titles/abstracts are passed through a dedicated sanitization function (_sanitize_text). This function neutralizes Markdown code-fence breakouts and YAML document delimiters, effectively mitigating risks of indirect prompt injection or data poisoning in generated files.
  • [SAFE]: Network communications are limited to trusted and well-known services, specifically Hugging Face (huggingface.co) and arXiv (arxiv.org), for legitimate paper indexing and metadata retrieval.
  • [SAFE]: Template rendering logic uses context-aware escaping (_escape_yaml_value) to ensure that user input cannot break the structure of YAML frontmatter in generated research articles.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 01:28 PM
Security Audit — agent-trust-hub — huggingface-paper-publisher