qemu-register-extraction
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is designed for hardware modeling and uses local resources for all operations. The Python validation script (
scripts/validate_register_ir.py) was analyzed and found to be safe, containing only file-parsing logic and regular expression validation without any network or dangerous subprocess capabilities.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it is designed to ingest external technical documents and source code. Evidence: (1) Ingestion points include driver source code, datasheets, firmware images, and SVD/IP-XACT files; (2) Boundary markers include instructions to summarize and filter for register-specific conclusions; (3) Capability inventory consists of workspace file writes and local script execution; (4) No automated sanitization of external text is implemented. This risk is inherent to the skill's primary purpose and is addressed by the structured audit and verification workflow.
Audit Metadata