Three tools with complete schemas and reasonable descriptions, but significant gaps in parameter documentation and error handling guidance. All tools are read-only (low risk), but descriptions lack actionable context for LLM selection. analyze_document and analyze_spreadsheet have similar names and overlapping functionality (both read files), risking LLM confusion. Parameter descriptions are minimal, 'path' lacks format guidance, 'max_chars' lacks rationale. No output schemas documented. Error handling is absent.

• Rename analyze_spreadsheet to summarize_spreadsheet_data or describe_spreadsheet_structure to disambiguate from analyze_document. Use verb_noun convention consistently.
• Document output schemas for all three tools in descriptions. Example: 'Returns {text: string, page_count: int, word_count: int, extracted_at: ISO8601}' so LLMs know what fields to expect.
• Add format constraints to 'path' parameter: 'Absolute or workspace-relative path. Supported: .pdf, .docx, .txt, .md. Max 100 MB.' This prevents invalid inputs and guides LLM behavior.
• Add error recovery guidance: 'If file not found, check path spelling and permissions. If file is too large, increase max_chars or split into smaller files.' This teaches the LLM how to self-correct.
• For generate_finance_report, document auto-detection fallback: 'If amount_column is omitted, the tool searches for numeric columns named amount, total, cost, or value. If none found, returns error with available columns.' This sets expectations.