brand-analyzer

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill maintains an attack surface where untrusted data provided by the user is processed and written into files.
  • Ingestion points: As described in Step 2 of the Core Workflow in SKILL.md, the skill collects external brand information, mission statements, and legacy materials from the user.
  • Boundary markers: The output templates in the assets/ directory (e.g., brand_guidelines_template.md) use standard placeholder syntax like {{MISSION_STATEMENT}} but do not include explicit instructions or delimiters to prevent the agent from following commands potentially embedded in that user data.
  • Capability inventory: The skill instructions guide the agent to perform file-write operations to the local filesystem, specifically creating markdown documents in the project root or a brand-documents/ directory.
  • Sanitization: The provided files do not contain logic for sanitizing or validating user-provided text before it is interpolated into the final reports and guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:07 PM
Security Audit — agent-trust-hub — brand-analyzer