a1-yandex-kit-promotions
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill proactively addresses indirect prompt injection risks by defining a trust boundary for external data. Ingestion points: Delivery notes, order comments, and product reviews. Boundary markers: Explicit 'Untrusted store text' instructions provided in SKILL.md. Capability inventory: Local Node scripts and API tools for store management. Sanitization: Instructions to ignore embedded commands and use only minimal excerpts.
- [COMMAND_EXECUTION]: Local execution is limited to search and validation scripts (search_docs.mjs). Analysis of the search script confirms it uses only Node.js built-ins and performs no unauthorized file or network operations.
- [EXTERNAL_DOWNLOADS]: The skill interacts with the official Yandex KIT API at api.kit.yandex.net. This is a well-known service endpoint necessary for the skill's stated purpose.
Audit Metadata