Server has 7 tools with explicit schemas and descriptions. Naming follows verb_noun convention (search_*, get_*, explore_*). Descriptions are present and moderately detailed (avg ~150 chars). However, several parameters lack descriptions, output schemas are not documented, and error handling is minimal. The search_drugs and search_companies tools include helpful enum descriptions and examples, but most other tools are sparse. No tool annotations (readOnlyHint, destructiveHint) despite all being read-only operations.

• Document output schemas for all tools. Example for get_drug: 'Returns {id: string, name: string, phase: string, company: string, indication: string, ...}'. This enables LLMs to plan downstream calls and extract the right fields.
• Expand get_drug, get_drug_swot, get_drug_financial descriptions to 80-150 chars each. Clarify what each returns and when to call it. E.g., 'get_drug_swot: Returns competitive strengths, weaknesses, opportunities, threats for a drug. Call after get_drug to analyze market position.'
• Improve explore_ontology description. Explain it searches taxonomy terms (not full drug records), returns matching terms with counts, and is useful for discovering valid values before calling search_drugs. Current description is too vague.
• Add error handling guidance to get_* tools. Document: 'If drug not found, try search_drugs() with partial name. If API rate limit hit, retry after 60 seconds.'
• Add readOnlyHint: true annotation to all 7 tools in the Tool definition. This signals to clients that these are safe to retry and cache.