DartLab exposes 7 financial analysis tools with Korean-language descriptions and basic input schemas. However, critical gaps undermine production readiness: (1) Descriptions are domain-specific but lack LLM-optimized guidance on WHEN to use each tool vs. alternatives. (2) Input schemas present but lack enums for constrained fields (e.g., 'group' in company_analysis accepts free-form strings like 'financial, valuation, governance' with no enum constraint). (3) No output schemas documented, LLMs cannot predict response structure for downstream chaining. (4) Parameter descriptions are minimal (e.g., 'stock_code' lacks format guidance; 'period' says 'if unspecified use latest' but no format specified). (5) No error handling guidance, tools fail silently on invalid inputs. (6) Tool names lack action verbs (company_analysis, company_panel, gather_flow are noun-heavy; should be analyze_company, get_company_panel, fetch_flow). Baseline: 194 chars avg description, 72 chars avg param annotation, these tools average ~100 chars description, ~40 chars param annotation.

• Rename tools with action verbs: 'analyze_company' (not 'company_analysis'), 'get_company_panel', 'fetch_flow', 'list_audit_risks', 'get_executive_compensation'. This signals intent to LLMs and disambiguates from similar tools.
• Add enum constraints to all categorical parameters. Replace 'group' description with enum: ['financial', 'valuation', 'governance', 'outlook', 'macro_sensitivity']. Replace 'topic' with enum: ['BS', 'IS', 'CF', 'audit', 'executive', 'business_overview'].
• Document output schemas for all tools. Specify: What fields does analyze_company return? What is the structure of company_panel response? Include sample JSON showing field names, types, and nesting. This enables LLMs to chain tools correctly.
• Expand parameter descriptions to 50-100 chars with format guidance. E.g., 'period: Fiscal year in YYYY format (e.g., 2024). If omitted, defaults to most recent available year.' For stock_code: 'KOSPI/KOSDAQ code (e.g., 005930) or EDGAR ticker (e.g., AAPL). Specify which market in the request context.'
• Add error handling and recovery guidance. E.g., 'If stock_code not found, call search_companies() with partial name. If date range is invalid, ensure start_date ≤ end_date in YYYY-MM-DD format. If period has no data, try the previous fiscal year.'