keystatic-astro
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents patterns for fetching and processing data from external APIs at build time, creating a surface for indirect prompt injection.
- Ingestion points: External data is ingested through the
fetch(apiUrl)method described inreferences/remote-collections.mdfor Astro content collections. - Boundary markers: The instructions mandate the use of Zod schemas (e.g.,
RemoteItemSchema.parse(data)) to validate the structure of the incoming data. - Capability inventory: The skill uses network operations (
fetch) and file system access (node:fs/promises) to retrieve and cache external content as JSON files within thesrc/content/directory. - Sanitization: While Zod validation ensures data types match, the skill does not include specific guidance for sanitizing the text content within those fields to prevent embedded instructions from influencing subsequent agent or build-time actions.
Audit Metadata