The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx @playwright/mcp@latest to download and execute the Playwright MCP server from the public NPM registry.
[COMMAND_EXECUTION]: Provides the playwright_evaluate tool, which executes arbitrary JavaScript within the browser's page context. While powerful, this execution is isolated to the browser sandbox.
[DATA_EXFILTRATION]: The playwright_screenshot tool allows writing files to the local filesystem via the path parameter, which could be used to export sensitive visual information from the browser to the host machine.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection because it ingests untrusted data from external websites that the agent processes.
Ingestion points: External web content is retrieved via playwright_navigate, playwright_snapshot, and playwright_evaluate in SKILL.md.
Boundary markers: No explicit delimiters are used to wrap or isolate ingested web content from the agent's system instructions.
Capability inventory: The agent has access to browser-side script execution (playwright_evaluate), file system writes (playwright_screenshot), and network navigation (playwright_navigate).
Sanitization: No specialized filtering or sanitization of web content is mentioned before it is added to the agent's context.

playwright-mcp-server