The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data including UX prompts, feature descriptions, product requirement documents, and design specs.
Ingestion points: The skill explicitly reads 'Product requirement docs, feature specs, UX handoff docs' and user-provided 'UX prompts' as primary inputs (Phase 1, Step 2).
Boundary markers: There are no instructions to use delimiters or warnings to ignore instructions embedded within these external documents.
Capability inventory: The agent has access to shell commands via rg (ripgrep) for file searching and npm/flutter for executing tests and analysis (Section Q).
Sanitization: The skill does not provide instructions for sanitizing or escaping the content retrieved from these external files before processing it.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to search the project and verify implementation.
Evidence: Section Q contains verification commands including rg, flutter analyze, flutter test, npm test, and npm run lint.
Context: These are standard development tools used for project analysis and quality assurance within the local development environment.
[DYNAMIC_EXECUTION]: The skill provides instructions for generating executable React, Flutter, and React Native components.
Evidence: Phase 2, Section H and Section R describe generating component code for handoff and implementation.
Context: This is a core function of the skill's purpose as a UI/UX designer and is considered low risk when generated from internal design logic, but could be influenced by the untrusted inputs mentioned in the indirect injection finding.

mobile-ui-ux-designer