Both tools have minimal descriptions (10-20 chars), no parameter descriptions, and no documented output schemas. Tool names are action verbs (web_search, extract_content) but descriptions are trivial. Input schemas exist but lack descriptions for parameters. No error handling guidance, no pagination support, and responses are JSON strings rather than structured objects. This server falls well below production baseline.

• Expand tool descriptions to 100-200 chars. Example: 'web_search: Search the web for information. Returns top results with titles, URLs, and snippets. Use when you need current information not in training data. Requires active billing credits.'
• Add parameter descriptions: 'query (string): Natural language search query. Can include operators like "site:example.com" or quoted phrases. Max 500 chars.'
• Document output schema explicitly. Declare that web_search returns {results: [{title, url, snippet}], total_results: number} and extract_content returns {content: string, url: string, extracted_at: ISO8601}.
• Add error recovery guidance: 'If payment_required=true, user must purchase credits at https://stinkmaster37.gumroad.com/l/mcp-credits before retrying.'
• Return structured objects (dicts/objects) instead of JSON strings. Let the MCP framework serialize; tools should return native types.