surprise-me

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface when using the 'News-based surprise' fallback. It instructs the agent to search for and ingest external news content, which could contain malicious instructions designed to influence the agent's behavior during the creative process.
  • Ingestion points: External data fetched via news search (SKILL.md, Step 3.1).
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified when processing external news data.
  • Capability inventory: The skill can read other skill files, execute their instructions, use search tools, and generate interactive HTML/React artifacts.
  • Sanitization: No sanitization, validation, or escaping of the ingested news content is described.
  • [DYNAMIC_EXECUTION]: The skill directs the agent to generate and execute complex artifacts including interactive HTML experiences, React components, mini-games, and generative art. While these are intended for creative deliverables, they involve the runtime assembly of executable code that incorporates user context and potentially untrusted external data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 11:14 PM
Security Audit — agent-trust-hub — surprise-me