Skills
skills/wanshuiyin/auto-claude-code-research-in-sleep/paper-illustration-image2/Gen Agent Trust Hub

paper-illustration-image2

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a project-local helper script at tools/paper_illustration_image2.py for preflight checks, finalization of artifacts, and verification. This follows a standard integration pattern for complex agent workflows on platforms like Claude Code.\n- [PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection by processing external user-provided text (e.g., paper abstracts or method descriptions) during the prompt engineering and review stages.\n
  • Ingestion points: User requests and paper documentation referenced in the SKILL.md workflow.\n
  • Boundary markers: Absent; the skill does not use specific delimiters or instructions to wrap user-provided data in prompts.\n
  • Capability inventory: Includes Bash for command execution and Write for file system access as described in the orchestration steps.\n
  • Sanitization: Not present; the skill summarizes user content but does not perform sanitization for embedded instructions before interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 06:06 PM
Security Audit — agent-trust-hub — paper-illustration-image2