operate-vmos-edge-cli
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's preflight instructions and the
scripts/ensure-installed.mjshelper script automate the installation of the@vmosedge/clipackage globally vianpm install -g. - [COMMAND_EXECUTION]: The skill uses
node:child_process(spawnSync) to execute system-level commands for environment verification (node --version,npm --version) and to run the vendor's CLI tool. The CLI tool itself provides adevice shellcommand to execute arbitrary shell commands on managed Android virtual devices. - [DYNAMIC_EXECUTION]: The skill allows for the execution of JavaScript expressions within the targeted Electron application using the
ui evalcommand. It also supports complex automation flows via YAML playbooks (vmos-edge-cli run) and JSON batch files (vmos-edge-cli batch). - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill reads untrusted data from the Electron UI state via
ui stateand diagnostic outputs from Android devices (e.g.,device info,device shell). - Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the data ingested from the UI or device outputs.
- Capability inventory: The skill has broad execution capabilities, including shell access on devices, UI interaction (click, type), and network request capturing.
- Sanitization: There are no explicit instructions for the agent to sanitize or escape data retrieved from the UI or devices before using it in subsequent prompts or logic.
Audit Metadata