app-intents-specialist
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEPROMPT_INJECTIONMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains explicit instructions to override the model's prior training and base knowledge with the provided context. Evidence in SKILL.md includes the statements 'This information unconditionally supersedes any prior training' and 'Supersedes prior training on these topics.'\n- [METADATA_POISONING]: The skill content claims the documentation was 'written and published by Apple,' which creates a metadata discrepancy with the actual author 'tartinerlabs.'\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and evaluate untrusted user-supplied source code, creating a surface for indirect instructions. 1. Ingestion points: User-provided Swift projects and App Intents code (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: None. 4. Sanitization: Absent.
Audit Metadata