The Agent Skills Directory

[NO_CODE]: The skill consists entirely of Markdown instructions, evaluation JSON, and metadata. There are no scripts (Python, JS, etc.) or binary files included that could execute arbitrary commands.
[SAFE]: The eval/eval.json file explicitly defines the skill as readOnly and specifies that it does not mutate external state. It uses the meta-ads MCP for data retrieval, which is a standard method for interacting with external APIs securely.
[PROMPT_INJECTION]: The instructions are designed to enforce a specific reporting structure and analytical methodology. There are no patterns detected that attempt to bypass safety filters or override system-level instructions in a malicious way. Phrases like 'No deviation' or 'Mandatory' are used to maintain output quality and structure for the diagnostic task.
[DATA_EXPOSURE]: The skill processes advertising data provided by the user (CSV, screenshots, or API). It does not request or access sensitive system files like SSH keys, AWS credentials, or environment variables. It specifically instructs the user on safe data management practices for marketing analysis.
[EXTERNAL_DOWNLOADS]: The skill mentions external tools and related skills (e.g., ad-to-landing-page-auditor) but does not perform automated downloads of untrusted code. The installation process uses the official goose-skills registry.

meta-ads-analyzer