This server has significant definition quality gaps. Both tools have basic descriptions and input schemas, but descriptions lack depth, parameters lack context about constraints/formats, output schemas are not documented, and error handling guidance is absent. The tools follow a verb_noun naming pattern (search_papers, extract_info) but descriptions are generic and do not explain when/why to use each tool or what distinguishes them. No parameter descriptions beyond a bare statement of what the parameter is. No evidence of output schema documentation. The definitions would not meet production standards for agentic tool quality.

• Add explicit output schemas to both tools. Example: search_papers should document: {papers: [{id: string, title: string, authors: [string], abstract: string, url: string}], total_count: number, limit: number}
• Expand tool descriptions to 50 - 200 chars. search_papers: 'Search arXiv for papers by topic. Returns up to 20 results with title, authors, abstract, and arXiv ID. Use extract_info to get full metadata for a specific paper. Requires valid topic keyword (e.g., "machine learning", "quantum computing").'
• Add parameter descriptions and constraints. For 'topic': 'Research keyword or phrase (e.g., "neural networks", "climate change"). 1 - 100 characters. Searches title, abstract, and author names.' For 'paper_id': 'arXiv identifier without 'arxiv:' prefix (e.g., "2301.12345" or "1234.5678"). Must match arXiv format YYMM.NNNNN or NNNNN.NNNNN.'
• Document pagination for search_papers. Add optional parameters: limit (1 - 100, default 20) and offset (0+, default 0). Return total_count and next_offset in response.
• Clarify tool chain. Explicitly state in search_papers description: 'The returned paper.id can be passed to extract_info() to retrieve additional metadata (keywords, publication date, citations).'
• Add error recovery guidance. Example: 'If search returns no results, try a simpler keyword. If paper_id is not found, use search_papers with the paper title to locate the correct ID.'
• Document external service constraints. Note: 'Searches are subject to arXiv rate limits. Typical response time is 1 - 5 seconds. Timeouts after 30 seconds return an error.'