qemu-model-verification
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill evaluates external, untrusted data sources such as QEMU logs, traces, and boot output to verify hardware model claims. This ingestion workflow represents an indirect prompt injection surface where adversarial content in the evidence files could attempt to manipulate the agent's logic or verdict.
- Ingestion points: Processing of artifacts in
.oh-my-qemu/subdirectories, includinglogs/andscripts/, as well as external trace files and DTBs. - Boundary markers: The skill uses highly structured templates (
claim-record-template.md) and a rigid reporting format which helps isolate data from the agent's instructional context. - Capability inventory: The agent has permissions to execute QEMU binaries, run local scripts, and interact with the Git environment (e.g., updating
info/exclude). - Sanitization: The 'Evidence provenance' and 'Failure classification' sections require the agent to verify data against the register contract and model source, acting as a verification layer against potentially malicious data.
Audit Metadata