ArcRift, developed by Eshaan Nair, is a local persistent memory layer that connects browser-based AI chats with local development environments. The app captures browser chat context and exposes it through a native Model Context Protocol server so local IDE agents can query past decisions and code snippets. It stores information in a local SQLite knowledge graph, performs semantic recall, and runs embeddings locally. Developers and AI power users who need continuous cross-session context benefit most.

ArcRift preserves session continuity between browser chats and IDE agents

ArcRift records and reuses conversational context by capturing browser chat data and making it queryable for local coding assistants. A browser extension collects chat content while a local process ingests entries into a SQLite knowledge graph, allowing IDE agents to retrieve previously shared decisions and code snippets to inform their suggestions during development workflows.

Recalled context is delivered via semantic search and local embeddings

The system exposes a recall pathway using semantic search and explicit 'recall_context' tools so agents can request relevant items from the graph. Embeddings are computed locally using Ollama before indexing, which means retrieved context is driven by semantic similarity rather than simple keyword lookup, and IDE agents query that indexed store for context-aware responses.

Installing and running ArcRift requires specific dependencies and a browser extension

ArcRift runs on desktop platforms and depends on a local toolchain. Supported systems include Windows (with a dedicated installer), macOS, and Linux. Required components and inputs include:

• Node.js to run the backend
• Ollama for local embedding processing
• a Chrome extension for real-time browser chat capture

The tool integrates into MCP-based workflows and targets privacy-focused developers

Because it implements a native Model Context Protocol server, ArcRift connects to MCP-compatible IDE agents such as Cursor, Claude Code, and Windsurf and supports major chat interfaces like Claude and ChatGPT. The project is open-source and described as privacy-centric and local-first, and it has attracted attention on GitHub within the MCP early adopter community, which helps with integration examples and community-driven improvements.

ArcRift is a pragmatic choice for developers seeking local continuity

ArcRift is a practical option for developers who need persistent local context across chat and coding sessions but who accept a hands-on setup and maintenance approach. It suits privacy-minded individuals integrating conversational assistants into editor workflows. Users preferring a turnkey, cloud-hosted experience may find its installation and configuration requirements a poor fit; plan time for setup and validation before relying on retrieved context in critical work.