This project provides a unified, self-hosted memory layer for AI tools, addressing the fragmentation of context across applications like Claude, ChatGPT, and Cursor. It enables users to store information once and recall it semantically across any connected AI client, ensuring data ownership and persistence independent of any single platform. The primary benefit is a seamless, context-aware AI interaction experience, eliminating the need for constant re-explanation.

How It Works

The system leverages Cloudflare Workers to create a personal, serverless memory backend. Data is stored and indexed using semantic search capabilities, allowing retrieval based on meaning rather than exact keywords. This approach utilizes Cloudflare's Vectorize for embeddings and D1 SQLite for structured data, all deployable on the free tier. The core advantage lies in its cross-tool compatibility via the Memory/Context Protocol (MCP), enabling a consistent context across diverse AI applications without vendor lock-in.

Quick Start & Requirements

Deployment to Cloudflare is streamlined via a one-click process, typically taking around two minutes. Users are prompted to set a secure AUTH_TOKEN for client authentication. Connecting AI clients involves either running a provided bash/PowerShell script for CLI tools (like Claude Code) or manually configuring custom instructions and MCP connectors within applications like ChatGPT and Claude (desktop/web). An openssl command is suggested for generating strong tokens. Detailed per-app connection steps are available in the project's wiki.

Highlighted Details

• Unified Memory: Stores and recalls context across multiple AI tools (Claude, ChatGPT, Cursor, etc.).
• Semantic Recall: Retrieves information based on meaning, not exact phrasing.
• Broad Capture Integrations: Supports saving context via CLI, Obsidian plugin, iOS shortcuts, browser extensions, and bookmarklets.
• OAuth 2.0 Support: Enables secure, token-less authentication for browser-based MCP clients.
• Self-Hosted & Free Tier: Runs within the user's Cloudflare account, designed for personal scale on the free tier.

Maintenance & Community

Development and community interaction are centered around GitHub Discussions, serving as the hub for release announcements, roadmap shaping, and user support.

Licensing & Compatibility

The project is released under the MIT License, permitting commercial use and integration with closed-source applications. It is explicitly designed to operate within Cloudflare's free tier limits for personal usage.

Limitations & Caveats

While designed for ease of use, initial setup requires familiarity with Cloudflare Workers deployment and specific client integrations. The OAuth flow necessitates a KV namespace (OAUTH_KV), which requires manual configuration if not automatically provisioned by the deployment script. The "personal scale" emphasis suggests potential performance or storage limitations for very high-volume enterprise use cases.