The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill enables an agent to interact with untrusted external web content while maintaining high-privilege capabilities, creating a significant attack surface for indirect prompt injection.\n
Ingestion points: The skill ingests data from web pages via Playwright's page.goto() and content locators (e.g., in tests/dashboard/load.spec.ts).\n
Boundary markers: There are no boundary markers or instructions to isolate untrusted page content from the agent's control logic.\n
Capability inventory: The deleteTestUser helper in tests/helpers/users.ts allows for administrative deletions via fetch using an ADMIN_TOKEN.\n
Sanitization: The skill lacks sanitization or validation of the web content it processes.\n- [External Downloads] (LOW): The skill instructs the user to install standard testing packages from the npm registry.\n
Evidence: npm install playwright @playwright/test is recommended in the setup section. These are widely used and trusted libraries from a reputable organization (Microsoft).\n- [Command Execution] (LOW): The skill uses the Playwright CLI to manage browser binaries and run test suites.\n
Evidence: Uses npx playwright test and npx playwright install for environment setup and operation.

qa-regression