DrissionPage MCP has 42 tools with significant quality gaps. Most tools lack proper English descriptions (many are Chinese), parameter descriptions are minimal or absent, and input schemas are incomplete or missing entirely. Tool naming is inconsistent (mix of verb_noun and noun patterns). No output schemas are documented. Error handling is minimal. The server appears to be an early-stage wrapper around the DrissionPage browser automation library with limited production-readiness.

• Translate all descriptions to English and expand to 50-200 characters. Include WHAT the tool does, WHEN to use it, and any prerequisites. Example: 'browser_quit' → 'Close the browser and release all resources. Call this when automation is complete or to reset state. Returns success message.'
• Document output schemas for all tools. Specify return type (string, object, array), field names, and types. Example for tab_list: 'Returns array of {id: string, title: string, url: string, active: boolean}'. This enables LLMs to chain tools.
• Add descriptions to all parameters. For 'id' fields, clarify format (CSS selector, element ID, tab ID). For 'type' enums, list valid values inline. Example: 'type: string (one of: text, value, index), determines how to match the option'.
• Add error handling guidance to destructive/irreversible tools (run_js, run_cdp, ele_remove_ele). Include recovery hints: 'If execution fails, check syntax with run_js_validate() first' or 'If element not found, call ele_get_interactive_elements() to refresh the list'.
• Standardize naming to verb_noun pattern. Rename page_* tools to get_page_* (get_page_html, get_page_title, get_page_url). This aligns with baseline patterns and reduces LLM confusion when selecting between similar tools.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry