This server exhibits significant quality gaps across naming, descriptions, and schema completeness. While tool names follow verb patterns (google_search, scrape_page, search_and_scrape), descriptions are present but minimal and lack context about when to use each tool versus alternatives. Input schemas are visible and properly typed, but parameter descriptions are sparse (e.g., 'Search query' for google_search.query lacks guidance on format, length, or language). Output schemas are completely undocumented, the codebase shows no return type definitions for any tool. The server is DEPRECATED (package.json explicitly marks it deprecated in favor of web-researcher-mcp), which raises immediate concerns about maintenance and current spec alignment. Error handling is not visible in the provided code. The server does implement tool instrumentation for metrics collection, but this does not offset the fundamental definition quality issues.

• Document the return type and structure for every tool. For google_search, explicitly state: 'Returns an array of result objects with fields: title (string), url (string), snippet (string). Maximum 10 results returned.'
• Expand parameter descriptions to 50 - 100 characters with format and constraint guidance. Example: google_search.query should be 'Search query (2 - 200 characters, any language, supports quoted phrases for exact matching)'.
• Split search_and_scrape into two separate tools: search_web (returns URLs and summaries) and scrape_url (fetches full page content). Agents can then compose them as needed.
• Add WHEN TO USE guidance to each tool description. Example for google_news_search: 'Use this to find recent news articles with dates. For web search results, use google_search instead. Includes freshness filtering (hour, day, week, month).'
• Implement structured error responses. When a tool fails, return { success: false, error: 'Search returned no results. Try simplifying the query or using different keywords.', retryable: true }. This guides LLM recovery.
• Consider adopting the web-researcher-mcp fork as the canonical implementation if this server is no longer actively maintained. Update the README to reflect the deprecation clearly.
• Add parameter constraints and enums where applicable. For sequential_search.action, verify the enum ['start', 'continue', 'conclude'] is enforced; consider adding a 'reset' action for state management.

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

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