The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches and analyzes untrusted data from external URLs such as ad libraries and customer reviews as part of its creative research workflow.
Ingestion points: Browser access to external sites described in references/creative-research-automation.md and references/google-ads-audit-checklist.md.
Boundary markers: The skill includes explicit security instructions in references/audit-guardrails.md and references/creative-research-automation.md to ignore embedded directives in fetched data.
Capability inventory: The skill uses MCP tools for ad platform management as noted in SKILL.md.
Sanitization: Clear instructions are provided to treat external content strictly as data and note any attempts to override instructions, effectively mitigating potential injection risks.
[EXTERNAL_DOWNLOADS]: The documentation refers to official tracking pixels from well-known services.
Evidence: Scripts for Google, Meta, and TikTok tracking in references/conversion-tracking.md.
Context: These references target official domains of established technology companies and are provided for user implementation, falling under the trust scope rule as safe.

ads