coach

Coach acts as a strategic partner for your development workflow by observing how you collaborate with Claude. Instead of offering generic advice, it parses your local session data to synthesize a precise profile of your thinking style, communication preferences, and primary technical domains. By tracking which suggestions you accept or decline, Coach becomes more effective over time, ensuring that the recommendations you receive evolve alongside your growing expertise. It identifies pain points such as repetitive tool misuse or inefficient prompting and matches them with targeted built-in skills, community marketplace plugins, or bespoke solutions generated via the skill-creator tool. This tool is ideal for developers who want to minimize friction in their coding sessions and maximize the utility of their AI assistant.

• Automatically analyzes historical session data through local /insights reports.

• Builds a persistent user profile tracking thinking style, pain points, and domain expertise.

• Filters out already installed skills to prevent redundant suggestions.

• Intelligently manages declined recommendations to deprioritize them in future sessions.

• Provides context-aware rationales for why a specific skill is relevant to your recent work.

• Requires Claude Code with /insights support and Python 3.8+.

• Keeps all data local; the history.json and usage reports never leave your machine.

• Integrates directly into your workflow via the /coach command.

• Best suited for power users of Claude Code who frequently switch between coding, debugging, and system architecture tasks.

• Uses beautifulsoup4 to parse raw HTML usage statistics into structured collaboration signals.