The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/with_server.py utilizes subprocess.Popen with shell=True to run commands provided via the --server argument. This capability is used to support complex shell commands for server startup but could be abused if inputs are untrusted.- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to visit and extract data from web pages.
Ingestion points: Data is ingested via page.content(), page.locator().all(), and screenshots as demonstrated in examples/element_discovery.py and recommended in SKILL.md.
Boundary markers: The skill does not implement boundary markers or instructions to isolate web content from agent instructions.
Capability inventory: The skill can execute shell commands through scripts/with_server.py and write to the local filesystem as seen in examples/console_logging.py.
Sanitization: Web content is not sanitized or validated before being presented to the agent for reconnaissance and action planning.

webapp-testing