The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external inputs such as testing requirements, behavioral specifications, and suggested locator changes from self-healing tools, which represents an indirect injection surface.
Ingestion points: Processes user-supplied testing requirements and automated locator proposals as described in the Instructions and Response Approach sections.
Boundary markers: Instructions mandate that the agent "Review changes against intended behavior and preserve original failures" and "Clarify goals, constraints, and required inputs."
Capability inventory: The skill utilizes various frameworks (Playwright, Selenium, Appium, K6) that involve subprocess execution, file system access for test implementation, and network operations for cloud testing.
Sanitization: The skill explicitly requires the use of "synthetic/approved data" and mandates separate authorization for cloud uploads, paid runs, and production writes.
[DYNAMIC_EXECUTION]: The skill facilitates the generation and execution of automated test scripts.
Evidence: Capabilities include "AI-driven test case generation," "self-healing test automation," and "Microsoft Playwright Code Generation." These behaviors are standard for the intended primary purpose of test automation engineering and are accompanied by instructions for manual verification and safety boundaries.

test-automator