The Agent Skills Directory

[Dynamic Code Execution]: The skill directs the agent to write and execute a Python script to perform supplier ranking calculations. \n
Evidence: The "Do this in code" section provides a template for a script that uses csv.DictReader to process local files. \n
Context: This is a routine pattern for ensuring the agent performs mathematical operations accurately rather than relying on qualitative reasoning. \n- [Data Ingestion Surface]: The skill processes data from external sources, which presents a surface for potential indirect prompt injection. \n
Ingestion points: Data is read from /mnt/user/data/supplier_catalog.csv and /mnt/user/data/suppliers.csv in SKILL.md. \n
Capability inventory: The agent uses Python script execution to handle this data. \n
Boundary markers: The instructions do not specify explicit delimiters or "ignore" warnings for the CSV content. \n
Sanitization: No explicit sanitization of the CSV fields is mentioned, though the logic focuses on arithmetic normalization. \n
Context: This ingestion is necessary for the skill's purpose. The risk is minimized by the structured nature of the data and the specific arithmetic operations requested.

supplier-selection