a1-yandex-kit-catalog-doctor
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a strict authorization gate requiring explicit user commands and verified value sources (e.g., owner-named exports or current API state) for any catalog modification.\n- [SAFE]: It enforces a multi-stage verification process where every mutation must be followed by a detail-read to confirm the result matches the intended state.\n- [SAFE]: Audit procedures are strictly read-only and use systematic pagination rules to ensure complete coverage without accidental data modification.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests external catalog data (titles, descriptions, characteristics). However, it mitigates potential injection risks by isolating this data from its operational logic through structured protocols and the requirement for explicit user authorization for any write action.
Audit Metadata