metadata-skills

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE]: The skills for ORCID and PlumX Metrics correctly instruct the user to handle API keys via environment variables (ORCID_ACCESS_TOKEN, ELSEVIER_API_KEY) rather than hardcoding them, following secure development practices.
  • [EXTERNAL_DOWNLOADS]: All external network operations target well-known and trusted scholarly infrastructure providers, including Crossref, DataCite, ORCID, Wikidata, OpenAlex, ROR, and Elsevier. These are standard services for academic research and metadata retrieval.
  • [INDIRECT_PROMPT_INJECTION]: Several skills (e.g., academic-paper-summarizer, crossref-api) involve fetching and processing external research data in JSON, BibTeX, or RIS formats. While this represents a data ingestion surface, the processing logic is limited to standard parsing and display, with no dangerous capabilities (like shell execution or file writes) performed on the external content.
  • [COMMAND_EXECUTION]: Code snippets provided are utility-focused (calculating h-index, querying APIs) and use standard libraries like requests and networkx. No arbitrary command execution or suspicious subprocess calls were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 09:57 PM
Security Audit — agent-trust-hub — metadata-skills