The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands such as grep, find, wc, and xargs within references/audit.md to analyze the project's source code. These commands are scoped to the project directories to identify hardcoded styling values and measure design system consistency.
[EXTERNAL_DOWNLOADS]: The skill references a feedback command npx --yes submit-expo-feedback@latest which downloads and executes a tool from the official NPM registry. This tool is a resource provided by the vendor ('expo') for reporting skill-related issues.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to audit and process project source files (TS/TSX), which introduces an attack surface where an agent reading these files might be influenced by instructions hidden in code comments or string literals.
Ingestion points: The auditing instructions in references/audit.md require the agent to read and evaluate the contents of arbitrary source files within the user's project.
Boundary markers: There are no explicit instructions or delimiters defined to separate code logic from potential instructions embedded within those files during the audit.
Capability inventory: The agent has the capability to create and modify theme files (src/theme/) and shared components based on its analysis of the project files.
Sanitization: The skill does not prescribe any sanitization or filtering of the file contents (such as ignoring comments) before the agent processes them.

expo-design-system