revyl-cli-dev-loop
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill extensively uses the
revylCLI to orchestrate development tasks, including starting dev loops, interacting with mobile devices (tap, type, swipe), and managing remote sessions. It also utilizesnpx expofor creating tunnels to the development environment. - [EXTERNAL_DOWNLOADS]: The instructions direct the agent to install and use external tools such as
expoandrevyl-cli-auth-bypassto facilitate the development and testing process. These tools are standard for the described mobile development workflow. - [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by processing data from the mobile application under test (e.g., screenshots and reports) and providing a natural language interface via commands like
revyl device instruction. * Ingestion points: Untrusted data enters the agent context through device screenshots (revyl device screenshot), device reports (revyl device report), and app content analyzed viarevyl device instruction. * Boundary markers: There are no explicit instructions for using delimiters or boundary markers when the agent processes content from the device. * Capability inventory: The skill has significant capabilities including executing shell commands (revyl), writing local files (--out), and performing network operations (--remote,--tunnel). * Sanitization: The skill does not define specific sanitization or validation protocols for data received from the external device session.
Audit Metadata