yandex-metrika

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow (SKILL.md "STOP! Перед любым анализом" steps) and scripts (e.g., scripts/common.sh metrika_get/metrika_get_csv and scripts/search_engines.sh, scripts/search_queries.md) fetch data from the public Yandex Metrika API (api-metrika.yandex.net) — including user-generated fields like search phrases, referrers and UTM values — which the agent is expected to read and analyze, so those untrusted third-party contents could indirectly inject instructions that influence subsequent actions.