The Agent Skills Directory

[SAFE]: The skill provides standard documentation and patterns for BI developers. It does not contain any hidden or malicious instructions.
[COMMAND_EXECUTION]: The skill uses a local script scripts/measure_check.py to perform verification of measure logic against user data. The script uses the pandas library to process data locally and does not perform any unsafe system operations, network requests, or arbitrary command execution.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided CSV data for calculation verification, which presents a standard attack surface for data-driven skills.
Ingestion points: User-supplied CSV files and JSON specifications are read by the measure_check.py script.
Boundary markers: The instruction set requires the agent to explicitly define the data model, grain, and evaluation context before generating any code or running tests.
Capability inventory: Generates BI formula code and executes a local Python verification script.
Sanitization: Data is processed via standard pandas.read_csv functions without additional sanitization, but the scope is limited to local numerical and date analysis.

bi-measure-builder