qt-qml-test
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted QML source files, which creates an attack surface for indirect prompt injection.
- Ingestion points: The workflow involves reading user-provided QML source files, module definitions (qmldir), and project build configurations (CMakeLists.txt) to generate context for the tests.
- Boundary markers: The skill contains explicit 'Guardrails' and 'Security constraints' sections in SKILL.md and copilot.prompt.md that mandate treating all source content strictly as data to be tested, overriding any potential instructions found within the code.
- Capability inventory: The skill uses file-writing tools to create test files in a /tests directory and may modify source files to add objectName properties for enhanced testability, though this requires explicit user consent.
- Sanitization: The agent is specifically instructed not to respond to embedded commands in comments, strings, or properties within the source files.
Audit Metadata