quicklook
Warn
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill incorrectly labels links to
sosumi.aias "Official Apple Documentation" in the References section ofSKILL.md. Official Apple API references are hosted ondeveloper.apple.com. Misrepresenting third-party sources as official is deceptive and can lead developers to trust unverified guidance. - [INDIRECT_PROMPT_INJECTION]: The skill describes patterns for processing and previewing external local files, creating a vulnerability surface where malicious file content could exploit system-level parsing frameworks.
- Ingestion points: The
fileURLandurlsparameters inSKILL.md(DocumentRowView, DocumentPreviewCoordinator) andreferences/quicklook-patterns.md(QuickLookSheet, ProgressiveThumbnailLoader) accept arbitrary local file paths. - Boundary markers: Absent. The skill does not include instructions to use delimiters or security warnings when handling content from untrusted local files.
- Capability inventory: The skill utilizes
QLPreviewControllerandQLThumbnailGenerator(referenced inSKILL.mdandreferences/quicklook-patterns.md), which involve complex system-level parsing and rendering of document, image, and 3D media content. - Sanitization: The
QuickLookValidatorclass inreferences/quicklook-patterns.mdprovides basic Uniform Type Identifier (UTType) validation to ensure files match supported formats, but no sanitization of the actual file data is performed.
Audit Metadata