a1-yandex-kit-launch-check

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a strong security posture by enforcing a read-only default for launch checks. It explicitly prohibits write operations like creation, updates, or deletions during the diagnostic phase.
  • [SAFE]: The 'Exact write-plan protocol' (references/exact-write-protocol.md) provides a high-quality safety framework for authorized mutations. It requires mandatory 'read-before-write' and 'read-after-write' verification steps to prevent unintended state changes or race conditions.
  • [DATA_EXPOSURE_&_EXFILTRATION]: The skill accesses the storefront URL (b2c_url) and performs HTTP status checks. This behavior is documented as a core function for verifying site availability and uses a controlled web-adapter boundary to mitigate risks associated with visiting untrusted URLs.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes data from external storefronts and API entities.
  • Ingestion points: Data retrieved from the store's public URL (b2c_url) and API responses for products/orders.
  • Boundary markers: Explicit instructions to treat the workflow as read-only by default and the use of the 'Authorization gate' for writes.
  • Capability inventory: Orchestrates write-capable skills (operator, catalog-doctor, promo-launcher) for updating store data.
  • Sanitization: Implements strict verification reads and minimal mutation logic (changing only authorized elements while preserving others) as described in the exact-write-protocol.md file.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 03:20 PM
Security Audit — agent-trust-hub — a1-yandex-kit-launch-check