Server has 6 tools with explicit schemas and descriptions. All tools follow verb_noun naming (search_posts, get_post, get_profile, etc.). Descriptions are present and adequate (100-200 chars), explaining WHAT each tool does. Input schemas use proper JSON Schema with type, minLength, maxLength, minimum, maximum constraints. However, parameter descriptions are minimal, most lack actionable guidance on format, dependencies, or error recovery. Output schemas are not documented in the source code. Error handling returns generic error results without recovery guidance. Tool annotations (readOnlyHint) are present, which is good. Missing: pagination guidance in descriptions, per-parameter descriptions for limit/query, and structured error messages that guide LLM recovery.

• Add per-parameter descriptions to all tools. For 'query' in search_posts, specify: 'Keyword query (1-500 Unicode chars). Supports natural language; wildcards not supported. Example: "machine learning"'. For 'limit', explain: 'Maximum results to return (1-50, default 10). Larger limits may increase latency.'
• Document output schemas explicitly. Create a schema comment block for each output type (PostOutput, RepliesOutput, ProfileOutput, etc.) listing all fields, types, and meanings. Example: 'Returns {post: {id, text, author, created_at, reply_count}, error: string|null}'.
• Enhance error handling with recovery guidance. Instead of generic errorResult(), return structured errors: {error: 'User not found', code: 'NOT_FOUND', suggestion: 'Try search_posts() with a partial username'}. Categorize as retryable|user_fixable|fatal.
• Add pagination guidance to tool descriptions. For tools with 'limit', add: 'Results are paginated. Call with limit=50 to fetch up to 50 items. No cursor/offset available; results are ranked by relevance.'
• Document URL format constraints. For 'post' parameter, specify: 'Canonical Threads post URL (e.g., https://www.threads.net/@username/post/123456789). Must be a valid public post URL; private posts return error.'