This server has significant quality gaps across naming, descriptions, schemas, and error handling. While tool names start with verbs (get_, start_, clear_, next_), descriptions lack sufficient detail for LLM decision-making, parameter descriptions are sparse, and output schemas are not documented. The server provides 5 tools with basic functionality but insufficient metadata for production use. Input schemas exist but are minimal, and descriptions are generic without actionable context about when/why to use each tool.

• Document output schema for every tool. For example, 'get_git_course_status' should return: {type: 'object', properties: {status: {type: 'string'}, lesson_idx: {type: 'integer'}, step_idx: {type: 'integer'}}} or similar. This enables LLMs to understand what data is available.
• Expand tool descriptions to 50-200 characters and include context. Example: 'clear_git_course_state: Resets course progress and deletes saved state. Use when user wants to start over. WARNING: Irreversible action.' This guides LLM selection and shows LLM that it modifies state.
• Add per-parameter validation rules and examples. For 'questions_answered' and 'insists' in next_git_course_step, document how the LLM should set these: 'questions_answered: true only if user explicitly states correct answers to all questions in the step. If unsure, set false and ask the user.'
• Implement comprehensive error handling with actionable recovery. When lesson_idx or step_index is out of bounds in 'start_git_lesson_step', return: 'Invalid lesson index 5. Available lessons: 0-3. Try another index or call get_git_course_status() to see current progress.'
• Rename 'start_git_lesson_step' to 'jump_to_lesson_step' to disambiguate from 'start_git_course' (which resumes, not jumps). This prevents LLM confusion about intent.
• Remove or complete the commented-out sampling code. If sampling is not supported, delete the comments entirely. If it will be added, clearly mark it as TODO with a target date. Incomplete code signals instability.