Playwright Tools for MCP - A Model Context Protocol server that provides browser automation tools using Playwright
The server exposes only 1 tool (browser_connect) with a minimal schema. The tool name lacks a clear action verb and the description is vague about what 'connect' accomplishes. The input schema is present but extremely sparse (only a 'method' enum). No output schema is documented. The codebase shows many additional tools (screenshot, navigate, evaluate, etc.) defined in src/tools.ts, but the sample provided shows only browser_connect registered. Without visibility into how these tools are actually exposed via the MCP protocol, they cannot be scored. The single visible tool falls well below production quality.
Connect to a browser context using one of the available methods
Tool name 'browser_connect' lacks a clear action verb. Should be 'launch_browser', 'start_browser', or 'open_browser' to signal intent. Current name is ambiguous about whether it launches a new browser or attaches to an existing one.
Description 'Connect to a browser context using one of the available methods' is too vague. Does not explain what happens after connection, what the tool returns, or when to use it. Lacks the WHAT/WHEN/RETURN structure required for LLM selection.
No output schema documented. LLMs cannot plan downstream calls without knowing what fields (e.g., browser_id, context_id, session_token) the tool returns.
Many tools defined in src/tools.ts (screenshot, navigate, evaluate, keyboard, mouse, etc.) but only browser_connect visible in the sample. Cannot assess whether these tools are properly registered with the MCP server or if they have schemas and descriptions.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | <=2025-11-25 | v2 |
Input parameter 'method' has an enum constraint but no description explaining what each browser type (chromium, firefox, webkit) is used for or when to choose one.