The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/with_server.py executes shell commands using the subprocess module to manage local development servers. It implements security controls including a restricted allowlist of permitted executables (npm, node, python, etc.) and validation to reject shell metacharacters and directory traversal attempts.
[DYNAMIC_EXECUTION]: The core workflow involves the agent dynamically creating and running Python scripts using the Playwright library to automate web interactions. This is the intended behavior for local environment testing.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and analyzing untrusted content from web applications (HTML, DOM structure, and rendered text), which could potentially contain malicious instructions intended to influence the agent's behavior. Ingestion points: Data enters the agent's context through Playwright functions such as page.content() and locator inspection as described in SKILL.md. Boundary markers: No specific delimiters or instructions to ignore embedded content are prescribed in the provided examples. Capability inventory: The agent has the ability to execute commands via the with_server.py script and manage local files, although these are scoped to the project environment. Sanitization: There is no evidence of content sanitization or filtering applied to the data retrieved from tested web applications.

webapp-testing