The x-reader server exposes 4 tools via FastMCP with functional but minimal descriptions. All tools are READ_ONLY with low risk. Tool naming is verb-forward and clear (read_url, read_batch, list_inbox, detect_platform). However, descriptions lack LLM-optimized detail about WHEN to use each tool, what distinguishes them from each other, and explicit parameter constraints. Input schemas are present for all tools but lack description text for parameters, particularly 'url' and 'urls' fields have minimal context about format validation, expected platforms, or failure modes. Output schemas are entirely undocumented: agents cannot see what fields read_url returns (title, content, url, source_type, platform metadata are mentioned in prose but not in a structured schema definition). No error handling guidance, no pagination info for list_inbox, no details on timeout behavior or partial failure modes for concurrent read_batch. STDIO transport caps the protocol readiness at 50 regardless of feature completeness.

• Add formal JSON Schema definitions for all tool outputs. Document read_url response as: {type: 'object', properties: {title: {type: 'string'}, content: {type: 'string'}, url: {type: 'string'}, source_type: {type: 'string', enum: ['youtube', 'bilibili', 'twitter', 'wechat', 'xhs', 'telegram', 'rss', 'generic']}, platform_metadata: {type: 'object'}}}
• Enhance 'url' parameter description: 'The URL to read (https://... or http://...). Supported platforms: YouTube, Bilibili, X/Twitter, WeChat, Xiaohongshu, Telegram, RSS feeds, or any generic HTTP(S) web page. Required format: absolute URL starting with http:// or https://. Invalid/unreachable URLs return a structured error with the reason (DNS failure, timeout, HTTP 404, etc).'
• Enhance 'urls' parameter description in read_batch: 'Array of URLs to fetch concurrently. Min: 1, Max: 100 per request. Timeout: 30 seconds per URL. Partially failed batches return mixed success/failure results in order. Example: ["https://example.com/a", "https://example.com/b"]'
• Add timeout and retry guidance to read_url and read_batch descriptions: 'Each URL fetch times out after 30 seconds. Network timeouts and temporary failures (HTTP 5xx) are retryable; permanent failures (4xx, invalid URL) are not. Agents should catch errors and either skip failed URLs or ask the user for alternatives.'
• Clarify list_inbox behavior and add optional pagination: Implement limit and offset parameters with description 'Paginate inbox results. Default limit=50, max=500. Returns items in reverse chronological order (newest first). Useful for long-running inboxes to avoid context window exhaustion.'