Server has 5 tools with complete JSON Schema definitions and descriptions. Naming follows verb_noun convention (replicate_*, get_*, list_*). Descriptions are present but brief (avg 95 chars, below 194-char baseline). All tools have typed input schemas with enums where appropriate (pageType, limit constraints). However, output schemas are inferred from TypeScript types rather than explicitly documented in tool registration. Error handling returns structured content with error codes and messages, but recovery guidance is minimal. No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels (WRITE vs READ_ONLY). Parameter descriptions are adequate but lack dependency hints and format guidance.

• Add explicit JSON Schema output definitions to each tool registration. Convert TypeScript types (ReplicationToolStructuredContent, etc.) to JSON Schema and include in tool definition so LLMs know what fields to expect.
• Add tool annotations: set destructiveHint=true on replicate_site_to_theme and replicate_site_to_hydrogen; set readOnlyHint=true on get_replication_job, list_replication_jobs, list_destination_stores.
• Expand tool descriptions to 150-250 chars. Include WHEN to use each tool and how it differs from similar tools. Example: 'Replicate a Shopify storefront to a Hydrogen target. Use this for headless commerce; use replicate_site_to_theme for theme-based replication.'
• Enhance error responses with recovery guidance. When replication_failed, include: 'Check job logs at [path]. Retry with different seedRoutes or contact support if stage X consistently fails.'
• Document parameter formats in descriptions. For seedRoutes, add: 'Array of URL paths (e.g., ["/products", "/collections/new"]). Max 25 routes. Omit domain; paths only.'
• Add pagination guidance to list_replication_jobs description: 'Returns up to 50 jobs. Use limit parameter to control batch size. No cursor pagination; use createdAt to fetch older jobs.'