eas-app-stores
Fail
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute unverified packages directly from the NPM registry using
npx. Specifically,npx testflightis used for TestFlight submissions andnpx --yes submit-expo-feedback@latestis recommended for sending feedback. These are not standard or official tools from Apple or Expo, and running unverified code from a remote registry poses a high risk of unauthorized system access. - [DYNAMIC_EXECUTION]: The skill documents and provides examples for using JavaScript-based configuration files (
store.config.js) for EAS Metadata. This allows for the execution of arbitrary JavaScript code, including network requests usingfetch(), during the metadata push process, which could be exploited to run malicious logic or exfiltrate environment data. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local project configuration files (
eas.json,app.json,store.config.json) and uses this data in shell commands and deployment tasks. - Ingestion Points: Configuration is read from JSON and JS files in the project root.
- Boundary Markers: The skill lacks explicit instructions or delimiters to prevent the agent from following malicious instructions embedded within these configuration files.
- Capability Inventory: The skill has the capability to run shell commands (
eas build,eas submit), write to external services, and access network resources. - Sanitization: There is no evidence of sanitization for values read from these configuration files before they are interpolated into commands.
- [EXTERNAL_DOWNLOADS]: The skill relies on external CLI tools and packages (
eas-cli,submit-expo-feedback) and provides instructions to install global packages and execute scripts from the NPM registry. The use ofnpxfor these tools bypasses version pinning and integrity checks, increasing the risk of supply chain attacks.
Recommendations
- AI detected serious security threats
Audit Metadata