CoLearner MCP exposes 36 browser automation and platform-specific tools with minimal documentation. Tool descriptions are present but extremely brief (avg ~15 chars), falling well below the 194-char baseline for production tools. No visible input schemas in provided source code, tool definitions appear to be inferred from tool.ts type definitions rather than explicitly registered with JSON Schema. Parameters lack type annotations and descriptions. Error handling is absent. The server is a fork of playwright-mcp with added WLXT/Overleaf integrations but inherits structural quality issues.

• Add explicit JSON Schema definitions for all 36 tools. Each tool must declare inputSchema with type, properties, required array, and per-parameter descriptions. Use zod-to-json-schema (already in dependencies) to generate schemas from TypeScript types.
• Expand tool descriptions to 50 - 200 characters. Include WHAT the tool does, WHEN to use it, and any prerequisites. Example: 'Navigate to a URL and wait for page load. Use this to start a new task or move to a different page. Returns page title and URL after navigation completes.'
• Add descriptions to every parameter. For 'selector' in browser_click, document: 'CSS selector or XPath to identify the element. E.g. "#submit-btn" or "//button[@type='submit']". Required.'
• Implement error handling with recovery guidance. Return structured errors like {code: 'ELEMENT_NOT_FOUND', message: 'Element not found. Try browser_wait_for_element first.', retryable: true}.
• Move credentials (username, password) from tool parameters to environment variables or a secure config file. Inject them server-side. Document in tool description: 'Logs in using credentials from server configuration. No parameters required.'