zog
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for processing untrusted data from external sources including HTTP requests, JSON payloads, and environment variables.
- Ingestion points: Data ingestion is handled by the
zhttp.Request,zjson.Decode, andzenv.NewDataProvideradapters as detailed inreferences/input-adapters.md. - Capability inventory: The skill enables validation, coercion, and transformation of ingested data via Zog schemas.
- Boundary markers: Schema definitions utilize explicit constraints like
.Required(),.Min(), and type-specific validators to define data boundaries. - Sanitization: Input data is processed through Zog's internal coercers and validation logic to ensure it matches the expected Go types and constraints.
Audit Metadata