eliasoulkadi/shokunin

Shokunin (the Japanese word for artisan) is a collection of 62 skill definition files for AI coding agents such as OpenCode, Claude Code, Cursor, and Windsurf, combined with a persistent memory system and supporting configuration. The core idea is to give an AI agent detailed, domain-specific knowledge organized into structured guides rather than generic instructions, each skill covers a specific engineering domain with decision-making logic, common error patterns, production checklists, and anti-patterns. The 62 skills span areas like Docker, Kubernetes, authentication, API design, frontend components, mobile development with Flutter and React Native, code review, SEO, PDF generation, and more. These are stored as markdown files that the AI agent reads to understand how to approach a given task. The memory system, built on ChromaDB (a local vector database), stores context from every coding session and allows the agent to recall relevant past decisions and information across sessions using a combination of three search strategies: vector similarity, BM25 text search, and temporal ranking. Everything runs locally with no cloud services required, so no data leaves your machine and there are no subscription costs. MCP servers (Model Context Protocol, a standard way to give AI agents access to tools) handle file access, web fetching, and memory integration. Installation is a single command on both Windows and Linux. You would use this if you want an AI coding assistant that feels consistent and knowledgeable across many sessions and projects, without paying per-session for cloud memory services.