NYC

pennylane

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE] (SAFE): The skill follows best practices for documenting an open-source library. No malicious instructions, hidden commands, or suspicious behaviors were identified across any of the files.
  • [EXTERNAL_DOWNLOADS] (SAFE): Installation instructions use uv pip to install established packages from PyPI, including pennylane and its official hardware plugins (e.g., pennylane-qiskit, pennylane-cirq). These are trusted scientific libraries.
  • [DYNAMIC_EXECUTION] (SAFE): The skill mentions @qml.qjit for just-in-time compilation. This is a standard performance feature of the PennyLane/Catalyst ecosystem for compiling quantum programs and does not constitute a security risk.
  • [INDIRECT_PROMPT_INJECTION] (SAFE): While the skill includes patterns for ingesting classical data into quantum circuits (e.g., angle_encoding), the risk is negligible as it involves numerical data processing within a scientific computing context. 1. Ingestion points: angle_encoding(x, wires) and classifier(x, weights). 2. Boundary markers: Absent. 3. Capability inventory: Python execution for quantum simulation. 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:20 PM