api-cms-sanity
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates data ingestion from an external source (Sanity Content Lake) which inherently creates a surface for indirect prompt injection if the source content is compromised. However, the instructions provide strong defensive patterns to mitigate these risks:
- Ingestion points: Data is primarily ingested through
client.fetchoperations as shown inexamples/core.md. - Boundary markers: The skill mandates the use of parameterized GROQ queries (
$param) and thedefineQuerywrapper to prevent injection attacks and ensure structured data handling. - Capability inventory: The skill documents capabilities for fetching data and performing document mutations (create, patch, delete) using the
@sanity/clientlibrary. - Sanitization: It emphasizes the use of Sanity TypeGen for end-to-end type safety and recommends structured rendering of rich content using Portable Text rather than raw HTML interpolation.
Audit Metadata