The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/with_server.py uses subprocess.Popen with shell=True to execute server commands provided via command-line arguments. It also uses subprocess.run to execute automation commands. While intended for server lifecycle management, this allows for arbitrary shell command execution.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to interact with and extract data from web applications, creating a vulnerability surface where malicious content on a web page could influence agent behavior.
Ingestion points: Data is ingested via page.content(), page.locator().all(), and browser console logs (page.on("console", ...)) as demonstrated in examples/element_discovery.py and examples/console_logging.py.
Boundary markers: There are no explicit instructions or delimiters to tell the agent to ignore or treat embedded instructions in the web content as data only.
Capability inventory: The skill possesses the ability to execute shell commands (via scripts/with_server.py) and write files (via examples/console_logging.py and examples/static_html_automation.py).
Sanitization: No sanitization or validation of the ingested web content is performed before the agent processes it.
[DYNAMIC_EXECUTION]: The skill's primary workflow involves the dynamic generation and execution of Python scripts using the Playwright library to perform UI automation and testing tasks.

webapp-testing