Konteks

Konteks is a memory engine for AI coding agents.

It builds a project-local context graph through autonomous knowledge curation, ensuring you never re-explain your project to an AI agent.

Memory artifacts are stored directly inside your repository, exposing compact, task-specific recall through an MCP server without requiring global installation or cloud dependencies.

🚀 Key Features

• Zero-Install: Run anywhere via npx or bunx without native dependencies.
• Language-Aware: Understands code structure across various file types.
• Local-First: Your project memory stays in your repo—no cloud, no accounts.
• Token-Efficient: High-fidelity context synthesis designed for LLM economy.

🧠 How It Works

Konteks treats a repository as a local memory system for agent workflows. It builds searchable project artifacts from source files, retrieves task-relevant evidence, and expands related context through a lightweight graph of entities and relations when available.

For a deep dive into the philosophy, architecture, and usage, see the Full Documentation.

• Overview: Vision, Philosophy, and the "Why."
• Session Lifecycle: How to work with Konteks (Warm Up -> Build -> Save).
• Architecture Overview: How the memory engine works under the hood.
• Glossary: Short definitions for Konteks terms.

⚡ Quickstart

Konteks runs on Node.js (>=22) or Bun. Start by initializing memory from your project root:

npx -y konteks-cli init

Continue with the Quickstart for MCP setup and the Warm Up -> Build -> Save flow.

📂 Local Storage

Konteks writes local memory under .konteks/. It uses SQLite (WASM) for the graph/indexes and a content-addressed object store for payloads. No host SQLite client or native modules are required.

⚖️ License

MIT Licensed. See LICENSE for details.