check-studio-skills

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill instructs the agent to fetch external data (documents of _type == 'skill') and treat the content field as authoritative guidance equivalent to the skill's own instructions. This represents an indirect prompt injection vulnerability surface.
  • Ingestion points: Data is ingested via the query_documents and get_document tools targeting a Sanity Studio dataset.
  • Boundary markers: The skill lacks explicit boundary markers or instructions for the agent to ignore potentially malicious commands embedded within the fetched data; instead, it explicitly tells the agent to follow them as 'authoritative guidance'.
  • Capability inventory: The skill uses document querying and retrieval tools. The primary risk is that injected instructions could influence how the agent uses other available tools or handles user data during prose generation.
  • Sanitization: There is no evidence of sanitization, validation, or filtering of the fetched content before it is applied to the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 10:33 PM
Security Audit — agent-trust-hub — check-studio-skills