The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the execution of npx --yes submit-expo-feedback@latest. This command downloads a package from the NPM registry and executes it immediately. The package source is not associated with a recognized trusted vendor or the primary Expo organization.
[EXTERNAL_DOWNLOADS]: The skill downloads several specific dependencies, including react-native-css@0.0.0-nightly.5ce6396 and nativewind@5.0.0-preview.2. These are specific nightly/preview builds from the NPM registry. The feedback submission tool also requires downloading an external package at runtime.
[COMMAND_EXECUTION]: The skill uses multiple shell commands (npx expo install, npx submit-expo-feedback) to install dependencies and configure the project environment.
[INDIRECT_PROMPT_INJECTION]: The feedback mechanism creates a vulnerability surface by interpolating user-provided or agent-generated strings into a shell command.
Ingestion points: The <actionable feedback> placeholder in the feedback command within SKILL.md.
Boundary markers: None. The input string is enclosed in double quotes but lacks validation or escaping against shell injection.
Capability inventory: The skill includes shell command execution capabilities via npx for installation and feedback reporting.
Sanitization: There is no evidence of sanitization or filtering for the feedback string, which could allow for command injection or exfiltration of sensitive information if the agent is manipulated.

expo-tailwind-setup-v55