The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs users to configure and reference highly sensitive credential files, specifically Apple App Store Connect API keys (.p8 files) and Google Service Account JSON keys. The requirement for the agent to interact with and configure these specific file paths in automated scripts creates a direct exposure vector for these credentials.
[EXTERNAL_DOWNLOADS]: The skill promotes the command 'npx testflight' as a primary method for build submission. This is not an official Expo or EAS command. Using 'npx' with an unverified package name allows for the execution of untrusted code from the public npm registry, which could be exploited through dependency confusion or malicious package publishing.
[DYNAMIC_EXECUTION]: The documentation for EAS Metadata highlights the use of 'store.config.js'. This format enables arbitrary JavaScript execution during the build and metadata sync process. The example provided includes performing network requests via 'fetch' to retrieve external data, which could be used to inject malicious content or exfiltrate environment variables.
[INDIRECT_PROMPT_INJECTION]: The skill manages application metadata, including titles and descriptions. As this data is often sourced from external inputs or generated by the agent, it creates a surface for indirect prompt injection attacks where malicious instructions could be embedded in the metadata processed by the deployment tools.
[COMMAND_EXECUTION]: The skill relies on numerous shell commands for deployment workflows. The combination of these commands with sensitive credential access and dynamic script execution increases the potential impact of any command injection or unauthorized execution.

expo-deployment