kodo

# kōdo (コード)

**Universal persistent memory for AI coding agents.**

【[English](README.md) | [中文](README.zh-CN.md)】

[![Claude Code](https://img.shields.io/badge/Claude_Code-compatible-orange)](https://docs.anthropic.com/en/docs/claude-code)
[![Cursor](https://img.shields.io/badge/Cursor-compatible-blue)](https://cursor.sh)
[![Kiro](https://img.shields.io/badge/Kiro-compatible-green)](https://kiro.dev)
[![Codex CLI](https://img.shields.io/badge/Codex_CLI-compatible-purple)](https://github.com/openai/codex)
[![MCP](https://img.shields.io/badge/MCP-server-red)](https://modelcontextprotocol.io)
[![MIT License](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)

> Your AI coding agent forgets everything between sessions. You tell it "use ESM imports" — next session, it's back to `require()`. You fix a bug caused by missing null checks — next session, same bug. **kōdo fixes that.**

An agent-agnostic memory layer that stores structured memories — conventions, mistakes, decisions, preferences, patterns — in a local SQLite database, and makes them available to **every** agent you use. One memory store. Every agent. Zero cloud.

```
kodo add -t convention -c "Always use ESM imports, never require()"
kodo add -t mistake    -c "Never use process.exit() in library code — throw instead"
kodo learn             # auto-learn from git history
kodo export            # sync to .cursor/ .cursor/ .kiro/ .codex/
```

## The Problem: AI's Amnesia

Every AI coding agent starts every session with a blank slate. This means:

| What happens | How it feels |
|-------------|-------------|
| You correct the agent's code style | It forgets next session |
| You explain your architecture decisions | Gone. Every. Time. |
| The agent makes a mistake you've seen before | Groundhog Day |
| You switch from Cursor to Claude Code | Start over from scratch |
| New team member onboards with an agent | Zero institutional knowledge |

**kōdo gives your AI agent a persistent, structured, searchabl