Critical failure: Tool definitions are inferred from agent.py comments, not explicitly registered with schemas. No input schemas visible in source. Tool descriptions are generic placeholders ("Search engine tool loaded from MCP server..."). No parameter documentation. No output schema documentation. No error handling guidance. This server does not meet minimum MCP tool definition standards.

• Create explicit tool registration with full JSON Schema. Define input parameters with types (string, number, enum), descriptions, and constraints. Example: search_engine should accept query (string, required), marketplaces (array of enum values), limit (integer, 1-50, default 20).
• Rewrite tool descriptions to answer: WHAT does it do? WHEN to use it? WHAT does it return? Target 50-200 characters. Example: 'Search for products across multiple marketplaces (Amazon, eBay, Allegro, etc.). Returns product name, price, marketplace, URL, and rating. Use this to find available items matching user query.'
• Document output schema with typed fields. search_engine should return: {results: [{product_name, price, marketplace, url, rating, product_id}], total_count, next_cursor}. web_data should return: {title, description, price, specifications, reviews, availability}.
• Add parameter descriptions for each input. Example: 'marketplaces: Array of marketplace names to search. Valid values: Amazon, eBay, Allegro, Kaufland, Morele, RTV Euro AGD, MediaExpert. At least one required.'
• Implement error handling with recovery guidance. Example: 'Marketplace unavailable. Try again in 30 seconds or select a different marketplace.' Include per-item success/failure for batch operations.