The Agent Skills Directory

Data File Interaction: The skill is designed to read multiple data files located in the /mnt/user/ directory, including stock_levels.csv, products.csv, sales_history.csv, and purchase_orders.jsonl. This interaction is the primary mechanism for inventory report generation.
Dynamic Script Generation: The instructions guide the agent to write and execute a single Python script to process large datasets efficiently. While this is a standard practice for data analysis tasks, it involves the dynamic generation and execution of code at runtime.
External Data Ingestion Surface: Because the skill processes data from external CSV and JSONL files, it creates a surface for indirect prompt injection. This is a common security consideration when an agent ingests and acts upon data that could potentially contain unstructured content or instructions from external sources.
Ingestion points: Data is read from files in /mnt/user/data/ and /mnt/user/sinks/ as specified in SKILL.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the data ingestion.
Capability inventory: The skill utilizes code execution (Python script generation) to process data as described in SKILL.md.
Sanitization: Explicit sanitization or validation logic for the ingested file content is not detailed in the instructions.

weekly-report