eas-update
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill downloads and runs CLI tools directly from NPM using
npx. This includeseas-clifor configuration and publishing, andsubmit-expo-feedbackfor reporting issues. - [EXTERNAL_DOWNLOADS]: Required packages and tools are downloaded from the official NPM registry. These are vendor-provided resources from Expo.
- [COMMAND_EXECUTION]: Several operations involve executing shell commands to modify project files, install the
expo-updatespackage, and perform EAS operations. - [INDIRECT_PROMPT_INJECTION]: The feedback tool represents an injection surface where user-provided text is included in a shell command. 1. Ingestion points: User feedback strings and agent-collected context in
SKILL.md. 2. Boundary markers: Uses double quotes for the shell command argument but lacks explicit sanitization or instructions to handle special shell characters. 3. Capability inventory: The skill utilizesBashfor variousnpxandeascommands. 4. Sanitization: No validation or sanitization of the input string is specified before it is interpolated into the feedback command.
Audit Metadata