The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to download and install Playwright browser binaries and system dependencies.
Evidence: pwsh -Command "playwright install --with-deps" in SKILL.md.
Note: This is a standard installation procedure for a tool from a well-known service (Microsoft).
[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for reading and asserting against content from external web pages, creating a surface for indirect prompt injection if the agent processes malicious data.
Ingestion points: page.TextContentAsync(), page.InnerTextAsync(), and page.Console event handling in SKILL.md.
Boundary markers: No delimiters or "ignore embedded instructions" warnings are provided in the examples.
Capability inventory: The skill utilizes page.EvaluateAsync(), page.ClickAsync(), page.FillAsync(), and page.RouteAsync(), which allow for significant interaction with the browser environment.
Sanitization: No sanitization or validation of external page content is demonstrated before it is used in assertions or logic.
[DYNAMIC_EXECUTION]: The skill utilizes Playwright's capability to execute arbitrary JavaScript within the browser context.
Evidence: await page.EvaluateAsync(...) is used for programmatic navigation and custom wait helpers in SKILL.md.
[COMMAND_EXECUTION]: The skill suggests command-line arguments and utilities that modify security settings for development and certificate management.
Evidence: The use of Args = new[] { "--ignore-certificate-errors" } in SKILL.md to bypass SSL validation.
Evidence: dotnet dev-certs https --export-path cert.pfx -p YourPassword in SKILL.md for certificate exportation.
[CREDENTIALS_UNSAFE]: The skill includes hardcoded dummy credentials in several code examples.
Evidence: await page.FillAsync("input[name='password']", "P@ssw0rd") and await page.GetByTestId("password-input").FillAsync("password123") in SKILL.md.

playwright-blazor