The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/with_server.py uses subprocess.Popen with shell=True to execute commands provided via the --server argument. This allows for the execution of arbitrary shell strings (e.g., cd backend && python server.py). The script also uses subprocess.run to execute the final test command.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to interact with and extract data from external websites, creating a surface for indirect prompt injection. A malicious site could include instructions in the DOM or console logs to trick the agent into executing dangerous commands via the included server helper.
Ingestion points: Example scripts like examples/element_discovery.py and examples/console_logging.py read DOM content (page.content(), page.inner_text()) and capture browser console messages.
Boundary markers: None. The instructions do not provide delimiters or warnings to the agent to ignore instructions found within the processed web content.
Capability inventory: The agent has the ability to execute arbitrary shell commands (via scripts/with_server.py), write files to /mnt/user-data/outputs/ (via examples/console_logging.py), and navigate the network using Playwright.
Sanitization: No sanitization or filtering is applied to the data extracted from the web browser before the agent processes it.
[DYNAMIC_EXECUTION]: The scripts/with_server.py utility acts as a dynamic execution wrapper, constructing and running shell commands based on runtime arguments provided by the user or the agent.

webapp-testing