bootstrap

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill generates a CLAUDE.md file containing instructions for the agent based on untrusted user input collected during the onboarding interview.
  • Ingestion points: Data enters the context via the AskUserQuestion tool during the interview stages and by reading existing files (BRAND.md, SOUL.md) during the migration phase.
  • Boundary markers: The skill does not define specific delimiters or 'ignore embedded instructions' warnings for the content written into the generated CLAUDE.md, strategy/brand.md, or about/me.md files.
  • Capability inventory: The skill utilizes file system tools to create directories, write multiple markdown files, and delete legacy files and folders.
  • Sanitization: No explicit sanitization, validation, or escaping of the user-provided content is performed before it is interpolated into the instruction templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 01:06 PM
Security Audit — agent-trust-hub — bootstrap