core-nfc
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [SAFE]: The skill documents standard iOS development practices for NFC integration, emphasizing correct entitlement usage (preferring
TAGover legacyNDEFvalues) and mandatory privacy keys such asNFCReaderUsageDescriptionin Info.plist. - [INDIRECT_PROMPT_INJECTION]: The skill describes an architecture that ingests data from physical NFC tags, which are untrusted external sources. 1. Ingestion points:
didDetectNDEFsanddidDetect tagsdelegate methods inSKILL.mdandreferences/nfc-patterns.md. 2. Boundary markers: None present in the code snippets for filtering tag content; however, the session availability is checked viareadingAvailable. 3. Capability inventory: Reading and writing NDEF messages, parsing payload records, and sending APDU commands to tags. 4. Sanitization: The examples demonstrate how to extract text and URLs but do not provide explicit sanitization logic for the extracted values before application use. - [METADATA_POISONING]: The metadata fields including name, description, and contents are consistent with the provided code snippets and do not contain deceptive instructions.
Audit Metadata