baoyu-cover-image

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill processes user-supplied article content (title, summary, keywords) and interpolates it into prompts for an image generation backend. A malicious article could contain instructions aimed at the downstream image model to bypass visual constraints or generate unintended content (Indirect Prompt Injection).
  • Ingestion points: Pasted article content saved to 'source.md' and interactive user input questions.
  • Boundary markers: The skill uses markdown headers in 'prompts/cover.md' but lacks explicit escaping or 'ignore instructions' delimiters for the interpolated content.
  • Capability inventory: The skill has access to tool invocation for image generation and shell execution of generation scripts.
  • Sanitization: There is no evidence of filtering or character escaping for article text before interpolation into the backend prompt.
  • [COMMAND_EXECUTION]: The skill facilitates the discovery and execution of local and sibling scripts (such as 'baoyu-image-gen' and 'baoyu-codex-imagegen') to perform image rendering. It uses the 'bun' runtime or 'npx' to execute these components, involving spawning sub-processes based on dynamically discovered paths in the project structure.
  • [EXTERNAL_DOWNLOADS]: The skill uses 'npx -y bun' as a fallback mechanism, which may download and execute the 'bun' runtime from the public NPM registry if it is not present in the execution environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 08:05 AM
Security Audit — agent-trust-hub — baoyu-cover-image