gsp-brand-identity

Pass

Audited by Gen Agent Trust Hub on Apr 11, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting content from multiple external project files and style presets into the creative director agent's context. \n
  • Ingestion points: Processes data from BRIEF.md, strategy chunks, mood boards, and style base configuration files. \n
  • Boundary markers: Lacks explicit delimiters or instruction isolation markers when inlining external content into the agent prompt. \n
  • Capability inventory: The agent can write to the brand's identity directory and trigger secondary domain skills for enrichment. \n
  • Sanitization: No explicit validation or filtering of the ingested markdown or YAML content is performed before it is passed to the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 11, 2026, 04:01 AM