100Critical
The skill executes shell commands and performs
Behavior Analysis
Description verified — behavior matches claim
Claims to do
iOS Polish: You are performing a final design quality pass on SwiftUI views. Your job is not just to check for correctness — it's to make the interface feel *crafted*. Every Apple-quality app shares certain traits: generous spacing, purposeful typography, restrained color, fluid motion, and invisible accessibility. This skill teaches you to see what's missing and fix it.
Actually does
This skill instructs the AI to perform a design quality review of SwiftUI views. It provides a structured 10-dimension checklist covering spacing, typography, color, animation, accessibility, and HIG compliance. The AI is guided to identify 'red flags' in the provided SwiftUI code and 'make the change' to fix them, following specific principles and a pre/post-polish assessment.
Skill Info
NameElvinOuyang/claude-skill-collection/ios-polish
Registrygithub
Version52c90f4772ba
PURLpkg:github/ElvinOuyang/claude-skill-collection@52c90f4772ba?skill=ios-polish
Stars1
Source
SHA-256771e46794663...
SHA-15d3d8759a083...
MD57293398ec456...
TLSHda72fa5b73c0...
ssdeep384:qbBb8aHc...
Size15,989 bytes
Install
This skill has been flagged as potentially malicious. Review the findings below before installing.
Claude CodeDocs
/plugin marketplace add ElvinOuyang/claude-skill-collection/plugin install ios-polish@ElvinOuyang/claude-skill-collectionclaude-marketplace
https://github.com/ElvinOuyang/claude-skill-collection
Skills.shDocs
npx skills add https://github.com/ElvinOuyang/claude-skill-collectionAssessments (3)
AI Agent Traps ↗Shell command execution function detectedcriticalcommand execution
Shell command execution function detected
100% confidenceline 54
system(
High-risk action without approval gate — financial or destructive operation bypassing human confirmationhighfinancial action
High-risk action without approval gate — financial or destructive operation bypassing human confirmation
100% confidenceline 131
delete without confirmation
Agent instructed to modify codelowcommand execution
The skill explicitly instructs the agent to 'make the change' after identifying areas for improvement. This implies the agent has write access to the codebase and the capability to modify files. While intended for benign design improvements, this capability could be abused if the agent were compromised or given malicious instructions.
80% confidenceline 34
Work through each dimension in order. For each one, note what you'd change and why, then make the change.
External references for RAG/knowledge basemediumsupply chain, rag poisoning
The skill refers to external Markdown files (e.g., `references/typography.md`) for detailed guidance. If these external files are not securely controlled or are loaded from untrusted sources, they could be modified to inject malicious instructions, poison the agent's knowledge base, or subtly alter its reasoning.
90% confidenceline 62
See `references/typography.md` for the full text style hierarchy and Dynamic Type testing guide.
Highly authoritative language to guide reasoninginfobiased language
The skill uses strong, prescriptive language ('NEVER', 'Red flags', 'Principles') to enforce specific design choices and best practices. While aligned with the skill's goal of enforcing HIG compliance, this is a form of semantic manipulation designed to heavily influence the agent's reasoning.
70% confidenceline 260
NEVER - Add decorative animations that serve no UX purpose
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/ElvinOuyang/claude-skill-collection/ios-polish)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/ElvinOuyang/claude-skill-collection/ios-polish"><img src="https://mondoo.com/ai-agent-security/api/badge/github/ElvinOuyang/claude-skill-collection/ios-polish.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/ElvinOuyang/claude-skill-collection/ios-polish.svgSecure your AI agents
Skills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.
- Continuous scanning of skills across all registries
- Policy enforcement before skills reach your agents
- Integration with your existing security workflow