Skills
skills/wanshuiyin/auto-claude-code-research-in-sleep/paper-illustration/Gen Agent Trust Hub

paper-illustration

Fail

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill includes instructions to run bash tools/install_aris.sh to download and install helper scripts from an external 'ARIS repo', which constitutes unverified remote code execution.
  • [COMMAND_EXECUTION]: User-provided input from $ARGUMENTS is directly interpolated into a shell command, creating a command injection vulnerability if the input contains shell metacharacters.
  • [COMMAND_EXECUTION]: The skill references local scripts (tools/install_aris.sh, tools/extract_paper_style.py) that are not provided in the skill source code, preventing security verification.
  • [EXTERNAL_DOWNLOADS]: The skill performs network requests to Google's Generative AI APIs (generativelanguage.googleapis.com) to facilitate figure generation.
  • [DATA_EXFILTRATION]: The skill reads data from user-specified local files or URLs and transmits the content to external AI models during the processing phase.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
May 18, 2026, 06:06 PM
Security Audit — agent-trust-hub — paper-illustration