The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches product information (titles, descriptions) from public Etsy URLs, which are then summarized for the user. This external content enters the agent's context and could contain malicious instructions.
Ingestion points: The productUrl parameter processed in scripts/etsy_product_detail.py fetches raw HTML/JSON data from the marketplace.
Boundary markers: SKILL.md provides instructions to "Summarize only returned facts" and "Preserve missing values as unknown," though these are not cryptographically enforced boundaries.
Capability inventory: The skill has network capabilities via urllib and file-write capabilities for caching and session data storage within the workspace.
Sanitization: No explicit content filtering or sanitization of prompt-like characters is performed on the product descriptions before they are returned to the agent context.
[DYNAMIC_EXECUTION]: The script scripts/etsy_product_detail.py modifies the Python search path at runtime to load shared utilities.
The functions get_api_url and _resolve_output_path use sys.path.insert(0, ...) to dynamically include a _shared directory relative to the script location. While used for internal vendor dependencies, this represents dynamic loading from a computed path.

ecommerce-etsy-product-detail