Parses natural-language shopping instructions and extracts structured search criteria: product type, must-have attributes (e.g., gluten-free, organic), numeric price constraints, brand/size, and quantity. The skill outputs a concise criteria object used to form search queries and to validate product listings.
Activate immediately when the agent receives a shopping-related instruction before running any search or click actions. It's useful for guided shopping assistants, e-commerce bots, or any workflow that must respect user constraints like price caps or dietary requirements.
parse_query.py) referenced in the instructions (check repo for the implementation).references/ folder for experiment context.{}
A query-parsing skill for the WebShop shopping benchmark that instructs agents to extract product type, attributes, and price constraints from natural language. The SKILL.md is clearly written with good examples, but the referenced parse_query.py script is absent from the bundle, making the core workflow non-functional. Purely static instructional content with no executable scripts.
No security issues whatsoever — purely instructional text. Main problem is missing script dependency that the skill instructs agents to run. Architecture is decent but incomplete due to the missing script. Usefulness is limited to WebShop scenarios.
