Three tools with complete input schemas and solid descriptions. All tools have verb_noun naming (decodo_scrape, decodo_google_search, decodo_amazon_product). Descriptions are 150-250 chars, within the 10-1024 baseline. All parameters have type definitions and descriptions. However, output schemas are completely undocumented, LLMs cannot plan downstream calls or extract structured data. No error handling guidance. The _apiKey parameter violates secret-injection pattern by appearing as a required tool parameter instead of server-side injection. decodo_amazon_product has a logical issue: url and asin are both optional but at least one is required, this dependency is not enforced in the schema (only _apiKey is marked required).

• Document output schemas for all three tools. Specify the structure of the response object, field names, types, and what data is returned on success. Example: 'Returns {content: string, status_code: number, url: string, ...}'
• Move _apiKey to server-side secret injection. Accept it via environment variable or vault at startup, not as a tool parameter. Update tool descriptions to remove _apiKey parameter and add a note: 'Authentication is configured server-side.'
• For decodo_amazon_product, enforce the url|asin constraint in the schema using oneOf or allOf, or update the required array to reflect the actual constraint. Document: 'Provide either url or asin, not both.'
• Add error handling guidance to all tool descriptions. Example: 'Returns error if the URL is unreachable or the API key is invalid. Retry with a different geo location if blocked.'
• For decodo_google_search, document result limits and pagination. Example: 'Returns up to 100 results per page. Use page_count to retrieve multiple pages. Results are capped at 10 pages (1000 results).' Add a limit parameter if needed.