Skills
Navigate
skills/aegntic/claudeism/multi-platform-architect/Gen Agent Trust Hub

multi-platform-architect

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection vulnerability surface identified in the component generation logic.\n
  • Ingestion points: The user_input string is processed in the MultiPlatformGenerator.generate_from_input method and the RequirementsAnalyzer class within tools.py.\n
  • Boundary markers: No delimiters or isolation instructions are implemented to separate user-provided content from the fixed template structures.\n
  • Capability inventory: The skill performs file system write operations using open().write() in examples.py and the CLI entry point of tools.py to save generated components.\n
  • Sanitization: There is a lack of escaping or validation for the description field in the templates (JS, Python, YAML, Markdown), which directly embeds the raw user_input. This allows for potential injection of malicious logic or metadata into the generated files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 08:13 AM