The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The references/prerequisites.md file contains instructions to automatically install the @playwright/test and @axe-core/playwright packages via NPM. It also facilitates the download of browser binaries for Chromium, Firefox, and WebKit using the npx playwright install command.
[COMMAND_EXECUTION]: The suite makes extensive use of shell commands to manage parallel execution of testing subagents, create local result directories, and interact with the playwright-cli and npx tools to conduct its testing lifecycle.
[REMOTE_CODE_EXECUTION]: The test-generate/SKILL.md skill analyzes the project's source code and generates new Playwright test files (.spec.ts) that are then executed via npx playwright test. Additionally, most skills utilize playwright-cli run-code to execute arbitrary JavaScript snippets within the browser context to perform validations and extract metrics.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. Untrusted data enters the agent context through playwright-cli open "$URL" (Ingestion points: multiple files including test-security/SKILL.md). The skill lacks explicit boundary markers or delimiters to ignore embedded instructions (Boundary markers: absent). The agent has broad capabilities including file-writing and shell execution (Capability inventory: test-generate/SKILL.md, test-all/SKILL.md). No sanitization or filtering of external content is performed before processing (Sanitization: absent).
[CREDENTIALS_UNSAFE]: The test-flows/SKILL.md skill facilitates authentication testing by accepting user credentials as arguments or environment variables. While the skill includes instructions to avoid hardcoding secrets, it still requires the agent to handle sensitive credentials in plaintext during the execution of shell commands.

playwright-ui-testing