Kemdicode Mcp

What is kemdiCode MCP?

kemdiCode MCP is a Model Context Protocol server that gives AI agents and IDE assistants access to 124 specialized tools for code analysis, generation, git operations, file management, AST-aware editing, project memory, cognition and self-improvement, multi-board kanban, and multi-agent coordination. It supports 7 LLM providers and is designed for developers who want persistent, intelligent AI-assisted development workflows.

How to use kemdiCode MCP?

Install globally via npm: npm install -g kemdicode-mcp. Then add the server to your AI IDE configuration (e.g., Claude Code, Cursor, KiroCode, or RooCode) using a JSON block like { "mcpServers": { "kemdicode-mcp": { "command": "kemdicode-mcp" } } }. Describe your goal in natural language; the AI agent automatically invokes the appropriate tools.

Key features of kemdiCode MCP

• 124 tools across code analysis, generation, git, file management, and more
• Cognition layer with 8 self-improvement tools for persistent AI memory
• Multi-board kanban for task and project management
• Multi-agent orchestration and coordination
• Multi-model consensus for architecture decisions
• AST-aware editing and project memory
• Supports 7 LLM providers (OpenAI, Anthropic, Google, etc.)

Use cases of kemdiCode MCP

• Review code for security issues before committing
• Fix bugs with AI assistance and cross-session error tracking
• Compare architecture decisions using multiple LLMs (e.g., event sourcing vs CRUD)
• Persist project decisions and context across sessions via write-memory/read-memory
• Distribute work among multiple AI agents (backend, frontend, QA) on a kanban board

FAQ from kemdiCode MCP

What is the cognition layer and how does it solve AI amnesia?

The cognition layer consists of 8 interconnected tools (decision-journal, confidence-tracker, mental-model, intent-tracker, error-pattern, self-critique, smart-handoff, context-budget) that write structured records to Redis. This allows the AI to remember decisions, errors, and context across sessions, so it can resume work without asking "where were we?".

Does kemdiCode MCP send my data to external services?

No. All cognition records and project data live in Redis with configurable TTL (default 7 days). Nothing is sent to external services. The agent writes its own memories locally as it works.

What are the runtime requirements for kemdiCode MCP?

It requires Bun ≥1.0 or Node.js ≥18, and TypeScript 5.0. Redis is optional but required for the cognition layer and multi-agent features. The package is installed via npm and is licensed under GPL-3.0.

How do I configure kemdiCode MCP in my IDE?

After installing the npm package globally, add it to your Claude Code config (e.g., claude mcp add kemdicode-mcp -- kemdicode-mcp) or include it in your IDE's MCP servers JSON (e.g., for Cursor, KiroCode, RooCode). Then instruct the agent to use kemdiCode MCP tools by adding a note to your project's CLAUDE.md or .cursorrules.

Can I use kemdiCode MCP with any AI agent?

It is designed for AI agents and IDE assistants that support the Model Context Protocol. The README specifically mentions compatibility with Claude Code, Cursor, KiroCode, and RooCode, but any MCP-compatible client can be used.