The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple sources, including the device's view hierarchy and the project's source code.
Ingestion Points: The agent uses user-maestro-inspect_view_hierarchy to read element text and attributes, and Grep/Glob tools to read files from the workspace during the research phases.
Boundary Markers: No explicit delimiters or instructions are provided to the agent to disregard or sanitize instructions found within the ingested data.
Capability Inventory: The agent has the ability to write files to the workspace and execute shell-level automation commands through the run_flow and run_flow_files tools.
Sanitization: No sanitization or validation of the ingested UI text or code content is mentioned before it is used to influence the agent's decision-making process while building the test steps.
[DYNAMIC_EXECUTION]: The skill's primary workflow involves generating executable YAML files for the Maestro automation framework and immediately executing them to verify behavior on a live device.
Evidence: The skill instructs the agent to scaffold a test file in Phase 4 and build steps iteratively in Phase 6, appending commands to a YAML file and running them via the run_flow tool to confirm they work.

maestro-create-test