The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/with_server.py uses subprocess.Popen with shell=True to execute arbitrary commands provided via the --server argument. This is intended to facilitate the lifecycle of local development servers but allows for general shell command execution within the environment.\n- [PROMPT_INJECTION]: The skill instructions and example scripts encourage the agent to inspect and interpret rendered DOM content (e.g., in SKILL.md and examples/element_discovery.py). This creates a surface for indirect prompt injection where malicious instructions embedded in a web page could influence the agent's actions.\n
Ingestion points: The agent ingests data from web pages using page.content(), page.locator().all(), and page.screenshot() as specified in SKILL.md.\n
Boundary markers: There are no explicit instructions or delimiters used to separate untrusted web content from the agent's core instruction set.\n
Capability inventory: The agent has the ability to write and execute Python scripts and run shell commands via the with_server.py utility.\n
Sanitization: No sanitization or filtering logic is present to validate or escape content retrieved from the browser before it is processed by the agent.

webapp-testing