The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/with_server.py utilizes subprocess.Popen with shell=True to execute server startup commands (e.g., npm run dev) and subprocess.run to execute automation scripts. This pattern is necessary for shell-based task chaining required in local development environments.
[PROMPT_INJECTION]: The skill facilitates indirect prompt injection by design, as it ingests and processes content from potentially untrusted web applications.
Ingestion points: The agent is instructed to read web page content via page.content(), examine DOM attributes in examples/element_discovery.py, and capture browser console logs in examples/console_logging.py.
Boundary markers: Absent. The instructions do not provide specific delimiters or warnings to separate untrusted web content from the agent's primary instructions.
Capability inventory: The skill environment allows for shell command execution via scripts/with_server.py, file system writes for logging and screenshots, and network navigation through Playwright.
Sanitization: None. Data retrieved from the browser is used directly to identify selectors and verify application behavior.
[DATA_EXFILTRATION]: Several example scripts (examples/console_logging.py, examples/static_html_automation.py) perform local data exposure by writing screenshots and log files to shared directories such as /tmp/ and /mnt/user-data/outputs/.

webapp-testing