The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to read sensitive platform credentials directly from the local file system (e.g., ~/.claude-marketing/brands/{slug}/profile.json). Accessing files that contain credentials for services like Google Ads, Meta, and CRMs is a high-risk operation that exposes sensitive authentication data to the agent's context.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection because it ingests untrusted data from marketing platforms and local configuration files.
Ingestion points: Loads brand guidelines from ~/.claude-marketing/brands/{slug}/guidelines/_manifest.json and fetches data from external MCPs (Google Analytics, Google Ads, Meta Ads, CRM, and Email).
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: Uses the google-looker-studio MCP to execute dashboard creation and configuration based on ingested data.
Sanitization: No sanitization, escaping, or validation logic is defined for the external data before it is interpolated into dashboard specifications and setup instructions.

live-dashboard