callkit
Warn
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [METADATA_POISONING]: The skill contains deceptive information, claiming to target non-existent software versions such as iOS 26 and Swift 6.3. It also references fictional frameworks like 'LiveCommunicationKit' in its evaluation criteria. This systemic misinformation can lead an agent to provide hallucinated or broken implementations.
- [EXTERNAL_DOWNLOADS]: The skill directs users and agents to documentation hosted on a suspicious, non-standard domain (sosumi.ai). These references mimic official Apple documentation structure but originate from an untrusted source, posing a risk of misinformation or redirection to malicious content.
- [INDIRECT_PROMPT_INJECTION]: The provided code patterns (specifically in references/callkit-patterns.md) demonstrate the ingestion of external PushKit payloads without implementing any sanitization or validation logic.
- Ingestion points: VoIPCallManager.reportIncoming and NotificationService.didReceive process data directly from CXCallUpdate and push notification userInfo dictionaries.
- Boundary markers: None are specified; the skill lacks delimiters or instructions to ignore embedded commands in the payload data.
- Capability inventory: The skill possesses capabilities for network interaction (connectToServer) and system UI manipulation through CallKit.
- Sanitization: Untrusted data from the decryptedPayload (such as callerName) is used directly in system notifications and call updates without escaping or filtering.
Audit Metadata