CodeSense has 6 tools with complete input schemas and descriptions, but quality is inconsistent. Tool names follow verb_noun convention (search_, generate_, investigate_, validate_, get_, list_) which is good. However, descriptions are generic and lack LLM-optimization guidance. Most critically, output schemas are completely undocumented, LLMs cannot infer what fields to expect from responses, breaking downstream tool chaining. Parameters lack detailed constraints (ranges, formats, dependencies). Error handling is absent from tool definitions. The server reads code and generates plans, but the interface doesn't guide agents on when to use each tool or how results chain together.

• Document output schemas for all 6 tools. Specify field names, types, and structure. Example: generate_plan returns {planId: string, steps: [{action: string, rationale: string}], estimatedTime: number}. This enables tool chaining.
• Expand tool descriptions to 80-150 chars with clear intent. Example: 'Generate a step-by-step implementation plan for a coding task. Use fast mode for simple changes, thorough for complex refactors. Returns planId for validation.' Include when to use vs similar tools.
• Add parameter descriptions for enums. Example: method enum should explain 'semantic: AI-powered search across code meaning; grep: regex pattern matching; structural: AST-based code structure queries.'
• Add min/max constraints to numeric parameters. Example: limit 1-100 (default 10), mode enum with explicit fast/thorough descriptions.
• Include error recovery hints in descriptions. Example: 'If no results found, try broader query or different method. If plan generation fails, ensure codebase is indexed via search_codebase first.'