AGENT LAB: SKILLS

fabric-lakehouse

Pass

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: SAFE
Full Analysis
  • Prompt Injection (SAFE): The skill content is purely educational and instructional regarding Microsoft Fabric Lakehouse. No patterns of behavioral override or jailbreak attempts were found.
  • Data Exposure & Exfiltration (SAFE): No hardcoded credentials, sensitive local file paths, or unauthorized network operations were detected. The documentation correctly references standard Lakehouse paths (e.g., 'Files/bronze/').
  • Obfuscation (SAFE): No encoded content, zero-width characters, or homoglyphs were identified in any of the files.
  • Unverifiable Dependencies & Remote Code Execution (SAFE): The skill does not perform package installations or execute remote scripts. All code snippets are static examples for Spark and SQL operations.
  • Privilege Escalation & Persistence (SAFE): No commands related to privilege escalation (e.g., sudo) or persistence (e.g., crontab, shell profiles) are present.
  • Indirect Prompt Injection (LOW/SAFE):
  • Ingestion points: The skill provides instructions for reading data from external sources (S3, GCS, ADLS Gen2) in references/pyspark.md.
  • Boundary markers: None explicitly defined in the snippets.
  • Capability inventory: Includes file-write, SQL execution, and table optimization capabilities in references/pyspark.md.
  • Sanitization: No specific sanitization logic is provided in the examples.
  • Assessment: While the skill describes an ingestion surface, this is the core intended purpose of a Lakehouse tool. No malicious exploitation of this surface is present within the skill code itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 18, 2026, 08:12 AM