IntelliStudy has 4 tools with basic schemas but critical gaps. All tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization). Input schemas are present but minimal: sessionId and intent parameters lack type constraints, enums, or validation guidance. No output schemas documented. Tool names lack action verbs (RAG_ASK, SUMMARIZE, QUIZ, TRANSLATE are acceptable but not verb_noun style). No error handling guidance, no parameter constraints, no pagination support. The codebase shows Spring Boot + MCP integration but tool definitions are inferred from McpAgentLayer.java rather than explicitly registered with full schemas. This is a proof-of-concept, not production-ready.

• Expand tool descriptions to 100-200 characters. Example for RAG_ASK: 'Answer user questions by searching the uploaded document. Returns the most relevant passage and a confidence score. Use this after uploading a document via the upload endpoint. Requires a valid sessionId from the upload response.'
• Document output schemas for all tools. Example: RAG_ASK returns {answer: string, source_page: number, confidence: 0.0-1.0}. SUMMARIZE returns {summary: string, word_count: number, key_points: string[]}.
• Add parameter constraints. For 'intent', add: 'A natural language question (1-500 characters). Examples: "What is the main topic?", "Summarize chapter 3."' For 'sessionId', add: 'UUID format (e.g., 550e8400-e29b-41d4-a716-446655440000). Obtained from the document upload response.'
• Rename TRANSLATE to TRANSLATE_TO_BANGLA or add a 'target_language' parameter with enum [bangla, english, ...] to clarify intent.
• Add error handling guidance. Example: 'If sessionId is invalid, call upload_document first. If intent is too long, split into multiple questions.'