The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation indicates that Playwright will be installed automatically if not present, which involves fetching dependencies from the official npm registry.
[COMMAND_EXECUTION]: The skill executes browser automation commands via Playwright to interact with local or remote web applications and performs file system writes when saving screenshots.
Evidence: captureScreenshot function in test-helper.js writes to the local file system.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes content from external or local web pages which may contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The agent navigates to and reads content from URLs using page.goto() as seen in SKILL.md examples.
Boundary markers: There are no explicit instructions or delimiters used to separate untrusted web content from the agent's internal logic or instructions.
Capability inventory: The skill can perform clicks, fill form fields, and capture screenshots within the browser context, as defined in SKILL.md and test-helper.js.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from the web pages before interaction occurs.

webapp-testing