Skip to content

Kagura AI

Kagura AI Logo

Universal AI Memory Platform

Own your memory. Bring it to every AI.

MCP-native memory infrastructure that connects Claude Desktop, ChatGPT, Gemini, and all your AI platforms with shared context and memory.

What is Kagura AI v4.0?

A universal memory layer that makes every AI remember your preferences, context, and history across all platforms.

Morning: ChatGPT helps you plan your day
         ↓ (remembers your preferences)

Afternoon: Claude Desktop writes code with you
           ↓ (knows your coding style)

Evening: Gemini analyzes your documents
         ↓ (recalls your project context)

One memory. Every AI.

Why Kagura AI?

For Individuals

  • 🔒 Privacy-first: Local storage or self-hosted
  • 🚫 No vendor lock-in: Complete data export anytime
  • 🧠 Smart recall: Vector search + Knowledge graph
  • 🌐 Universal: Works with Claude, ChatGPT, Gemini, Cursor, Cline

For Developers

  • 💻 MCP-native: 31 tools via Model Context Protocol
  • 🔌 Easy integration: kagura mcp install for Claude Desktop
  • 🛠️ REST API: FastAPI server with OpenAPI
  • 📦 Production-ready: Docker, authentication, monitoring

For Teams (Coming Soon)

  • 👥 Shared knowledge: Team-wide memory
  • 🔐 Enterprise features: SSO, BYOK, audit logs
  • 📈 Analytics: Track team AI usage patterns

Core Features

1. Universal Memory

Store once, access from any AI:

# Via MCP tool (works in Claude Desktop, ChatGPT, etc.)
memory_store(
    user_id="jfk",
    agent_name="global",
    key="coding_style",
    value="Always use type hints in Python",
    scope="persistent",
    tags='["python", "best-practices"]'
)

2. MCP Integration

Claude Desktop (local, all 31 tools): 

kagura mcp install  # Auto-configure
# All tools available: memory, files, web, shell, etc.

ChatGPT Connector (remote, 24 safe tools): 

docker compose up -d
# Connect ChatGPT to http://localhost:8000/mcp
# Safe tools only (no file ops, no shell)

3. Knowledge Graph

Track relationships and patterns: - AI-User interaction history - Memory relationships - Learning patterns analysis - Topic clustering

4. Complete Data Portability

# Export everything
kagura memory export --output ./backup

# Import anywhere
kagura memory import --input ./backup

Quick Start

Option 1: Claude Desktop User

pip install kagura-ai[full]
kagura mcp install
# Restart Claude Desktop - Done!

Claude Desktop Setup →

Option 2: ChatGPT User

docker compose up -d
# Configure ChatGPT Connector: http://localhost:8000/mcp

ChatGPT Connector Setup →

Option 3: Self-Hosted Production

git clone https://github.com/JFK/kagura-ai.git
cd kagura-ai
cp .env.example .env  # Configure DOMAIN, POSTGRES_PASSWORD
docker compose -f docker-compose.prod.yml up -d

Self-Hosting Guide →

Available Tools (MCP)

Memory (6 tools): - memory_store, memory_recall, memory_search - memory_list, memory_delete, memory_feedback

Graph (3 tools): - memory_record_interaction - memory_get_related - memory_get_user_pattern

Web/API (10+ tools): - web_search, web_scrape - youtube_summarize, get_youtube_transcript - brave_web_search, fact_check_claim

File Operations (local only): - file_read, file_write, dir_list

System: - shell_exec (local only) - telemetry_stats, telemetry_cost

Documentation

Community

Status: v4.0.0 (Phase C Complete)

Recently Completed: - ✅ Phase A: MCP-First Foundation - ✅ Phase B: Graph Memory - ✅ Phase C: Remote MCP Server + Export/Import

Features: - ✅ 31 MCP tools - ✅ REST API (FastAPI) - ✅ MCP over HTTP/SSE (ChatGPT Connector) - ✅ API Key authentication - ✅ Memory export/import (JSONL) - ✅ Production Docker setup

Coming Next: v4.0.0 stable release

Built with ❤️ for universal AI memory