Three tools with complete input schemas and descriptions, but descriptions are generic and lack LLM-optimized guidance. All parameters have types and descriptions, meeting baseline requirements. However, output schemas are undocumented (no return type structure specified), and error handling lacks recovery guidance. Tool names follow verb_noun convention (firecrawl_search, firecrawl_scrape, firecrawl_crawl) but are somewhat generic. No tool annotations (readOnlyHint, destructiveHint) despite all being read-only operations. Descriptions average ~120 chars, within acceptable range but lack context on when to use each tool vs. alternatives.

• Document return types explicitly. E.g., firecrawl_search returns List[{url: string, title: string, description: string}]. firecrawl_scrape returns {markdown: string, html: string, metadata: {...}}. This enables LLM to extract and chain results.
• Add tool-specific descriptions explaining use cases: 'Use firecrawl_search to find URLs matching a query. Use firecrawl_scrape to extract content from a single URL. Use firecrawl_crawl to discover and extract content from multiple pages on a domain.'
• Add readOnlyHint: true to all three tools in schema registration to signal safety and enable aggressive retry logic.
• Enhance error messages with recovery guidance: 'Invalid URL format. Ensure URL starts with http:// or https://. Try firecrawl_search() to find valid URLs first.' or 'API rate limit exceeded. Retry after 60 seconds.'
• Add pagination to firecrawl_search: accept offset/limit parameters and return {results: [...], total: N, has_more: bool} to support iterating large result sets.

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

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