ArcRift
Persistent local memory layer for AI. ArcRift uses a extension and a native MCP server to sync context and decisions from your browser chats (Claude, ChatGPT, D
What is ArcRift?
Persistent local memory layer for AI. ArcRift uses a extension and a native MCP server to sync context and decisions from your browser chats (Claude, ChatGPT, DeepSeek) straight to your local IDE agents (Cursor, Claude Code, Windsurf) powered by a local SQLite knowledge graph.
Overview of ArcRift
What it is
ArcRift is a local-first memory layer for AI coding tools. It captures conversations from browser-based AI chats and makes that context available in local IDE agents. The project consists of a browser extension and a native MCP server, backed by a local SQLite knowledge graph. It is open source under the MIT license.
Key features
- Syncs context and decisions between browser chats (Claude, ChatGPT, Gemini, DeepSeek, Grok, Copilot, Mistral) and local IDE agents (Claude Code, Cursor, Windsurf, Claude Desktop).
- Builds a searchable knowledge graph from conversations, with sentence-level embeddings generated locally via Ollama.
- Automatically injects relevant context into new prompts before they are sent.
- Runs entirely locally with no cloud or subscriptions.
- Supports concurrent read/write via SQLite WAL mode.
Use cases
- Developers who want to avoid re-explaining project context across AI chat sessions.
- Teams that need consistent decisions and architecture knowledge persisted across tools.
- Users who prefer local-first data storage for privacy.
Getting started
One-command setup: bash npx arcrift-setup For the desktop app, download the installer from the Releases page. For the MCP server, build the backend and generate config: bash cd backend npm install npm run build npm run mcp:config See the repository README for detailed extension and MCP setup instructions.
License and pricing
License: MIT. Pricing: open source (no pricing information stated).
Based on the project's public README and repository metadata.