The Agent Skills Directory

[METADATA_POISONING]: The skill provides external links to sosumi.ai while incorrectly labeling them as official "Apple docs". This is a deceptive practice as sosumi.ai is a third-party domain and not an official Apple documentation source (developer.apple.com).\n- [METADATA_POISONING]: The skill consistently references a fictional "iOS 26+" version throughout its instructions and evaluation cases. This metadata poisoning could lead an AI agent to hallucinate non-existent API features or apply logic intended for a non-existent operating system version.\n- [EXTERNAL_DOWNLOADS]: The skill directs the AI agent to documentation hosted on sosumi.ai, which is not a recognized trusted vendor or well-known service.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided data models to generate chart visualizations. This creates a surface for indirect prompt injection via malicious chart labels or data values, as the skill lacks explicit boundary markers or sanitization logic. Ingestion points: User-defined chart data in SKILL.md and referenced files. Boundary markers: Absent. Capability inventory: UI rendering only. Sanitization: Absent.

swift-charts