The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute az account get-access-token for authentication and the fab tool to retrieve workspace identifiers. These operations are required for normal interaction with the Microsoft Fabric environment.\n- [EXTERNAL_DOWNLOADS]: The skill performs network requests to api.fabric.microsoft.com to communicate with the Livy API for session management and code submission.\n- [REMOTE_CODE_EXECUTION]: A core feature of the skill is submitting PySpark and Python code to be executed on remote Spark compute clusters.\n- [PROMPT_INJECTION]: The skill processes data from lakehouse tables, which introduces a surface for indirect prompt injection if those tables contain malicious instructions.\n
Ingestion points: Lakehouse data is read using spark.sql queries within the Spark session.\n
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the code templates.\n
Capability inventory: The skill has the ability to execute shell commands locally and arbitrary code on the remote Spark cluster.\n
Sanitization: There is no evidence of sanitization or validation performed on the data retrieved from external tables.

executing-spark