pqc-implementation-attack

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads research-grade cryptographic libraries and side-channel analysis tools from well-known GitHub repositories, including liboqs, PQClean, pqm4, and dudect. These downloads are documented as necessary components for a cryptographic laboratory environment.
  • [DYNAMIC_EXECUTION]: The skill generates and runs Python scripts for specialized tasks such as statistical handshake timing (timing_probe.py) and data entropy mapping for ransomware triage. These scripts are constructed from visible templates in the instructions to facilitate automated measurements.
  • [INDIRECT_PROMPT_INJECTION]: The skill involves processing external artifacts such as firmware images and ransomware samples. The instructions incorporate security best practices by explicitly recommending isolated analysis environments and chain-of-custody procedures for handling malware and incident data.
  • [COMMAND_EXECUTION]: The skill makes extensive use of system utilities including openssl, binwalk, strings, nm, and git to perform fingerprinting and forensic analysis of target PQC implementations. These operations are consistent with the skill's stated purpose of implementation-layer auditing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 04:36 PM