Three read-only tools with clear verb-noun naming (list_chapters, get_chapter, get_verse). All have descriptions (194 - 250 chars, within baseline 34 - 392 range) and proper JSON schemas with typed parameters. However, parameter descriptions are minimal or absent (e.g., 'Chapter number, 1-18' lacks context on what happens with out-of-range values). No output schema documentation, LLMs cannot predict response structure. No error handling guidance (what if chapter 25 is requested?). Missing pagination for list_chapters despite potentially returning 18 items. No tool annotations (readOnlyHint) despite all being read-only. Composition is sound (each tool has one responsibility), but descriptions could be more LLM-optimized per pattern:tool-description.

• Add explicit output schema documentation for each tool. Example for get_verse: 'Returns {chapter: number, verse: number, sanskrit: string, transliteration: string, translations: {sivananda, purohit, gambirananda, adidevananda}, commentary: string}'.
• Enhance parameter descriptions with validation rules and error cases. Example: 'chapter: Chapter number (1 - 18, integer). Returns error if out of range or if chapter data is unavailable.'
• Implement error handling that returns actionable messages. Example: 'Chapter 25 not found. Bhagavad Gita has 18 chapters. Try chapter 1 - 18.' instead of raw API errors.
• Add readOnlyHint: true to all three tool definitions to signal to clients that these are safe, cacheable, and non-destructive.
• Document response field types and nesting in tool descriptions or via a separate schema reference, enabling LLMs to plan downstream operations.