The Agent Skills Directory

[SAFE]: No security issues were detected. The skill correctly follows isolation protocols and focuses on its intended purpose of advertising optimization. All external references target well-known and trusted services in the digital marketing ecosystem, such as Meta, Shopify, and standard advertising analytics tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as store URLs and brand history logs, which is a necessary part of its advertising optimization function.
Ingestion points: The agent ingests external data from store_url and ad_account_status via user input, as well as historical optimization data from learnings.jsonl.
Boundary markers: The provided instructions do not specify explicit delimiters or boundary markers for this external data.
Capability inventory: The skill is limited to generating advertising strategies and copy; it does not possess capabilities for arbitrary command execution (subprocess/exec), file system modification (beyond structured logging), or network exfiltration.
Sanitization: No specific sanitization methods for external data are detailed in the instructions. While this constitutes a potential ingestion surface for indirect prompt injection, the skill lacks dangerous capabilities that would make such a vector exploitable. The severity is assessed as safe as this behavior is inherent to the primary function of the skill.

afa-fb