maestro-flows
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill documentation includes the standard installation command for the Maestro CLI, which downloads and executes a script from the tool's official domain:
curl -Ls "https://get.maestro.mobile.dev" | bash. This pattern is used in both the setup guide and the CI integration examples. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to generate YAML automation flows based on user-provided application requirements, which introduces a potential surface for indirect injection if untrusted requirements are processed.
- Ingestion points: The agent ingests user-defined test requirements and application descriptions to generate
.maestro/*.yamlfiles. - Boundary markers: No specific boundary markers or "ignore embedded instructions" warnings are suggested for the input data.
- Capability inventory: Generated flows can execute JavaScript logic (
evalScript), perform UI interactions (inputText,tapOn), and navigate the application. - Sanitization: The documentation does not specify sanitization procedures for user-provided data interpolated into the generated YAML flows.
- [DYNAMIC_EXECUTION]: The skill demonstrates the use of Maestro's
evalScriptfeature, which allows for the execution of JavaScript snippets within the automation environment to handle complex conditions, timestamping, and state management.
Audit Metadata