The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/with_server.py uses subprocess.Popen with shell=True to execute commands passed via the --server argument. This allows for arbitrary shell command execution on the host system.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted data from external web pages and browser console logs. Ingestion points: The skill explicitly instructs reading page content via page.content() (in SKILL.md) and capturing console messages (in examples/console_logging.py). Boundary markers: No delimiters or instructions are used to distinguish untrusted web content from the agent's internal instructions. Capability inventory: The skill environment provides access to shell execution via subprocess.run and subprocess.Popen in the provided scripts. Sanitization: There is no evidence of sanitization, escaping, or filtering of the fetched content or logs before they are handled by the agent.

webapp-testing