The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a feedback mechanism that uses npx --yes submit-expo-feedback@latest. This command downloads and executes a package from the NPM registry. The package is associated with the skill's author ('expo') and is used for a legitimate purpose of submitting actionable feedback.
[COMMAND_EXECUTION]: The skill instructs the agent to use standard search tools like find and grep to inspect local project files for compatibility checking. It also includes instructions for creating GitHub issues using gh issue create --repo expo/expo to report missing functionality, which is a standard developer workflow and includes an explicit requirement to seek user approval before posting.
[REMOTE_CODE_EXECUTION]: The use of npx to run the feedback tool involves executing remote code. However, the source is the official NPM registry and the package name aligns with the vendor's naming conventions, representing normal functionality for this skill's ecosystem.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted data (source code and configuration files) and has capabilities like network interaction. (1) Ingestion points: The skill reads Swift, JavaScript, and TypeScript source files, along with package manifests. (2) Boundary markers: The instructions do not specify explicit delimiters to ignore instructions embedded in the processed code files. (3) Capability inventory: The skill can execute shell commands, run NPM packages, and interact with GitHub via CLI. (4) Sanitization: There are no explicit instructions for sanitizing content read from the source files before it is used in CLI commands.

expo-migrate-module