The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves ad-analytics recipes and specialized scripts from the goose-skills repository using the gooseworks fetch command to perform performance diagnostics and market intelligence.
[COMMAND_EXECUTION]: Ad generation, brand management, and feedback recording are performed via the mcp__gooseworks__* suite of tools. The skill also utilizes the gooseworks CLI for credit checks and searching for analysis recipes.
[DYNAMIC_EXECUTION]: Instructions specify fetching analysis scripts into /tmp/gooseworks-scripts/ and executing them to perform complex tasks like lead quality analysis and competitor ad library scraping.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external brand websites and user-provided descriptions, creating a potential surface for indirect injection.
Ingestion points: Brand website content retrieved during the research phase (SKILL.md).
Boundary markers: None explicitly defined to separate untrusted website content from agent instructions.
Capability inventory: MCP tool execution, file system writes to agent-config/brands/, and network operations for submitting ad batches (SKILL.md).
Sanitization: The skill employs a reconciliation workflow that requires user confirmation before updating the brand kit with new information.

goose-ads