lazy-load

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local shell commands to perform framework detection and codebase analysis.
  • Evidence: Employs grep, find, and wc to inventory project files and cat heredocs to generate implementation files (e.g., LazyImage.jsx, NextLazyImage.jsx, LazyImage.vue) in the .claude/lazy-loading directory.
  • [PROMPT_INJECTION]: Indirect prompt injection surface identified as the skill processes local source files.
  • Ingestion points: Reads package.json and project source files (*.jsx, *.tsx, *.vue, etc.) via grep and find to count existing lazy-loading occurrences.
  • Boundary markers: No explicit delimiters used when reading file contents for inventory counts.
  • Capability inventory: Executes Bash shell commands and writes files to the local file system.
  • Sanitization: Employs strict matching for framework detection and uses single-quoted heredocs ('REACT', 'VUE', etc.) in Bash to prevent unintended variable expansion in the generated code templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 03:53 AM
Security Audit — agent-trust-hub — lazy-load