The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/with_server.py script utilizes subprocess.Popen(shell=True) to execute server start commands (e.g., 'npm run dev') provided via CLI arguments. It also uses subprocess.run to execute the subsequent test command. This is a standard functional requirement for a testing utility that manages the lifecycle of development servers and test runners.
[INDIRECT_PROMPT_INJECTION]: The skill provides templates and examples for testing voice-based interfaces ('voice-interface.test.ts'), which ingest natural language commands. This represents a potential surface for indirect prompt injection. Ingestion points: Voice command strings processed by the 'VoiceAssistant' component. Boundary markers: None identified in the instructional templates. Capability inventory: Shell command execution ('scripts/with_server.py') and browser automation ('playwright'). Sanitization: The provided examples do not include explicit sanitization or filtering for voice command inputs.
[DATA_EXPOSURE]: Several example scripts ('examples/console_logging.py', 'examples/element_discovery.py') write automation logs and screenshots to '/mnt/user-data/outputs/' and '/tmp/'. These are standard output paths for agent environments and do not indicate unauthorized data exfiltration.

webapp-testing