qiskit

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the qiskit and matplotlib packages via the uv package manager. These are well-known, legitimate libraries in the quantum computing ecosystem.
  • [METADATA_POISONING]: The skill's YAML frontmatter includes a risk: critical label. This is misleading for a technical instruction set and does not accurately reflect the safe nature of the provided commands, which could cause misinterpretation by automated scanners or users.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of user-supplied quantum circuit logic and execution parameters.
  • Ingestion points: User prompts for circuit design, optimization levels, and backend selection are incorporated into the workflow scripts.
  • Boundary markers: No explicit delimiters or instructions are present to prevent the agent from interpreting instructions embedded within user-provided circuit parameters or code comments.
  • Capability inventory: The skill includes commands for package installation and network requests to IBM Quantum Runtime APIs.
  • Sanitization: No input validation or sanitization is demonstrated for the user-supplied data used in the code snippets.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 03:37 AM
Security Audit — agent-trust-hub — qiskit