mobile-eas-dev
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and act upon data from the mobile simulator's screen via the
agent-device snapshotcommand. - Ingestion points: Data enters the agent's context through the output of
agent-device snapshotas described inSKILL.md. - Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded within the simulator UI content.
- Capability inventory: The skill has the capability to execute shell scripts (
apps/mobile/scripts/eas-dev.sh), source environment files (.env.eas-simulator), and interact with the device viaagent-devicecommands (press, fill). - Sanitization: There is no mention of sanitization, filtering, or validation of the data retrieved from the simulator screen before it is used to determine subsequent actions.
- [COMMAND_EXECUTION]: The skill relies on the execution of a local shell script
apps/mobile/scripts/eas-dev.shto manage the EAS cloud iOS simulator lifecycle, including installation, starting sessions, and reloading the app. - [EXTERNAL_DOWNLOADS]: The instructions recommend the global installation of the
agent-devicepackage usingnpm install -g agent-device. This tool, which is associated with the vendor's ecosystem, is used to drive the simulator but is not pinned to a specific version in the provided instructions.
Audit Metadata