a1-yandex-kit-operator
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Yandex KIT API (including order details, catalog variants, and promotion rules) and has the capability to perform write operations (such as confirming orders or updating SKU prices), creating a potential surface for indirect prompt injection.
- Ingestion points: The skill instructions in
SKILL.mddescribe reading data vialist_orders,list_variants,list_products,list_discounts, andlist_webhooks. - Boundary markers: The skill employs an 'Authorization gate' and strict target resolution rules (defined in
references/exact-write-protocol.md) to ensure that only intended objects are modified. - Capability inventory: The skill utilizes write-capable tools including
confirm_order,cancel_order,update_variant,validate_webhook, andupdate_webhook. - Sanitization: The
exact-write-protocol.mdmandates the use of minimal mutations, preservation of untouched fields, and post-write verification to validate state changes.
Audit Metadata