xyo-knowledge
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides technical guidance for building on the XYO Protocol. It identifies the Datalake as an untrusted source and requires Zod-based validation for all ingested payloads to mitigate injection risks.
- [INDIRECT_PROMPT_INJECTION]: The skill describes an interface for processing untrusted data from the XYO Datalake. (1) Ingestion points: Data is retrieved from the Datalake via RestDataLakeViewer or chain gateways. (2) Boundary markers: The skill explicitly labels external payload content as untrusted until verified. (3) Capability inventory: Modules include signing (Account/Wallet), storage (Archivists), and query logic (Diviners). (4) Sanitization: The skill mandates Zod validation (e.g., zodIsFactory) to check schema and shape before processing.
- [EXTERNAL_DOWNLOADS]: The instructions reference official, vendor-owned NPM packages (@xyo-network/*) for type definitions and API details.
- [SAFE]: Cryptographic identity management is documented with security best practices, including warnings about derivation paths and the use of secure placeholders.
Audit Metadata