ml-paper-writing

Pass

Audited by Gen Agent Trust Hub on Oct 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the retrieval of official conference assets. For instance, the NeurIPS template includes a Makefile that downloads style packages from media.neurips.cc using curl.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and draft papers based on content from a user's research repository. Reading untrusted local files (logs, configs, READMEs) to determine contributions constitutes an attack surface for instructions embedded in data, though the risk is low and inherent to the tool's primary purpose.
  • [COMMAND_EXECUTION]: Instructions within the skill guide the user to execute shell commands like find, grep, and ls for repository exploration. It also provides Python snippets utilizing the requests library to interact with academic metadata APIs for citation verification.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 1, 2026, 07:50 AM
Security Audit — agent-trust-hub — ml-paper-writing