skills/xiaomimimo/mimo-code/arxiv/Gen Agent Trust Hub

arxiv

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill makes network requests to legitimate, well-known academic services including export.arxiv.org and api.semanticscholar.org to fetch paper metadata and citation data.
  • [COMMAND_EXECUTION]: The skill uses a bundled Python script (scripts/arxiv.py) which relies exclusively on the Python standard library. It does not invoke shell commands or external binaries.
  • [DATA_EXFILTRATION]: No data exfiltration patterns were detected. The script fetches data into the local environment for analysis and does not transmit local sensitive information to external servers.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data (abstracts, titles) from arXiv. While this presents a theoretical attack surface common to all web-reading tools, the script performs basic whitespace sanitization and lacks high-privilege capabilities that would be susceptible to exploitation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 02:43 PM