blob-storage
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides instructional material for using Val Town's official storage API. All module imports and documentation links point to trusted vendor domains (esm.town and val.town). The instructions include a specific security rule to use environment variables for secrets rather than persistent blob storage.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data through tools like readBlob and API methods like blob.getJSON (SKILL.md). This presents an indirect prompt injection surface where stored content could influence agent behavior. Ingestion points include readBlob and blob.get; boundary markers are established via val-scoping versus account-global imports; capability inventory includes blob.set, blob.move, and deleteBlob; and no specific sanitization of the stored content is described. This is an inherent risk of data storage documentation and is managed via proper scoping advice.
Audit Metadata