The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill is designed for data transformation within a controlled local workspace environment.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface by processing external, potentially untrusted content from extract.md and strategy.json into its context.
Ingestion points: Reads from workspace/brand/extract.md and workspace/pages/[page-name]/strategy.json.
Boundary markers: Not explicitly defined in the provided instructions.
Capability inventory: Restricted to reading and writing local files; no network or command execution capabilities are utilized.
Sanitization: The skill structures data into specific Markdown and JSON formats, reducing the risk of downstream misinterpretation.

brand-profile