Server has 4 tools with clear verb-noun naming (delete_cell, execute_cell, execute_code, insert_cell). All tools have descriptions (avg 95 chars) and input schemas with typed parameters. However, descriptions are terse (under 150 chars) and lack context on when to use each tool or what they return. Parameter descriptions are present but minimal. No output schemas documented. Error handling guidance is absent. Missing tool annotations (destructiveHint for delete_cell, idempotentHint for execute_cell). Schemas are present but lack detail on return types and structure.

• Document output schemas for all tools. E.g., execute_cell should return {execution_result: string, output: string, error: string | null, execution_time: number}. This enables LLMs to chain tools and extract results.
• Expand tool descriptions to 100-200 chars explaining WHAT the tool does, WHEN to use it vs similar tools, and WHAT it returns. E.g., 'Execute a single cell in the active notebook kernel. Use this for running code in sequence; use execute_code for ad-hoc IPython commands. Returns execution output, errors, and timing.'
• Add tool annotations: mark delete_cell with destructiveHint=true; mark execute_cell and execute_code with idempotentHint=false (execution has side effects); mark insert_cell with destructiveHint=false but note it modifies notebook structure.
• Implement a dry-run or confirmation step for delete_cell. E.g., add a 'confirm' boolean parameter (default false) that returns what would be deleted without actually deleting. This prevents accidental data loss.
• Enhance parameter descriptions with constraints and examples. E.g., 'cell_index: 0-based index of the cell to execute (0 to notebook.cell_count-1)'; 'timeout_seconds: max execution time in seconds (1-3600, default 60)'; 'stream: if true, emit progress updates every progress_interval seconds for long-running cells'.