a1-yandex-kit-promo-launcher
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill demonstrates high security awareness by incorporating a dedicated 'Exact write-plan protocol'. This protocol ensures that no writes are performed without unambiguous authorization, complete target resolution, and post-mutation verification steps.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided data to configure promotions, which is an inherent surface for indirect injection. However, it effectively mitigates this risk through a comprehensive validation and authorization framework.
- Ingestion points: Processes user-supplied business conditions, dates, and item identifiers (SKILL.md).
- Boundary markers: Explicitly requires the agent to stop and ask for clarification if inputs are ambiguous and never infers critical values like time zones or limits (SKILL.md).
- Capability inventory: Utilizes MCP tools to interact with the Yandex KIT API for discount, promocode, and gift management.
- Sanitization: Mandates pre-execution reads to verify the existence and status of every variant, category, and collection target before writing (references/exact-write-protocol.md).
Audit Metadata