The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains instructions to execute an external package named submit-expo-feedback via npx. This package is not hosted under an official organization namespace (e.g., @expo/) and is requested with the @latest tag, which allows for the dynamic download and execution of unverified third-party code.
[EXTERNAL_DOWNLOADS]: The skill utilizes a local utility script scripts/fetch.js to retrieve documentation and schema files from the Expo service's official domains. While the intended use targets reputable sources, the script is a general-purpose HTTP client that allows the agent to fetch content from any arbitrary URL, which could facilitate the ingestion of untrusted data.
[INDIRECT_PROMPT_INJECTION]: The agent is instructed to fetch and follow content from external JSON schemas and Markdown documentation to generate and validate workflow files.
Ingestion points: The agent retrieves resources from api.expo.dev and github.com/expo using the fetch.js script to guide its output (SKILL.md).
Boundary markers: The skill lacks explicit delimiters or instructions to the agent to treat the fetched content as untrusted data or to ignore potentially malicious embedded instructions.
Capability inventory: The skill has access to file system operations (Read/Write) and shell command execution via Node.js and npx for various CLI tools.
Sanitization: There is no evidence of sanitization or validation performed on the fetched external content before it is used to influence the agent's logic or file generation.

expo-cicd-workflows-v55