How do I share context between AI coding tools?

Install ctxovrflw — a local daemon that connects to all your AI tools via MCP (Model Context Protocol). Run 'curl -fsSL ctxovrflw.dev/install.sh | sh' then 'ctxovrflw init'. It auto-detects your AI tools and configures shared memory. What you tell Cursor, Claude Code instantly knows.

Does ctxovrflw work with Cursor?

Yes. ctxovrflw works with Cursor, Claude Code, Cline, Windsurf, Claude Desktop, Copilot CLI, Gemini CLI, and 12+ more AI agents. It auto-configures MCP connections during setup.

Is my data sent to the cloud?

By default, no. ctxovrflw runs entirely locally — your memories stay on your machine. Cloud sync is optional, E2E encrypted (AES-256-GCM, Argon2id key derivation), and uses ephemeral 24-hour cloud storage. The cloud never sees your plaintext data.

How is ctxovrflw different from .cursorrules or project knowledge files?

Project files like .cursorrules are static, tool-specific, and don't sync across tools. ctxovrflw is dynamic shared memory — it learns as you work, syncs across all MCP-compatible tools, supports semantic search, and works across devices. It's memory that grows with you.

What is MCP (Model Context Protocol)?

MCP is an open standard for connecting AI agents to external tools and data sources. ctxovrflw speaks MCP natively, meaning any MCP-compatible AI agent can read from and write to your shared memory without custom integration.

What is shared agent memory?

Shared agent memory is a persistent memory layer that multiple AI coding agents can read from and write to. ctxovrflw provides shared agent memory via a local MCP memory server — when you tell Cursor something, Claude Code instantly knows it. All your AI tools share the same knowledge base.

Is ctxovrflw a local vector database?

Yes. ctxovrflw is a local vector database built on SQLite with ONNX embeddings for semantic search. It stores memories as vectors on your machine, enabling sub-millisecond similarity search without cloud dependency. It's optimized specifically for AI agent memory.

ctxovrflw — Shared memory for every AI agent

ctxovrflw is a universal AI context and memory layer — a local-first daemon that shares context across 19+ AI agents (Cursor, Claude Code, Cline, Windsurf, Claude Desktop, Copilot, and more) via MCP (Model Context Protocol). What you tell one AI, every AI knows.

How to share context between AI tools

Install ctxovrflw with one command: curl -fsSL ctxovrflw.dev/install.sh | sh. Run ctxovrflw init to auto-detect and configure your AI tools. Memories stored by any connected agent are instantly available to all others.

Features

Semantic search — finds memories by meaning, not just keywords, using local ONNX embeddings
MCP native — speaks the Model Context Protocol, connects to any compatible agent
E2E encrypted sync — AES-256-GCM encryption, zero-knowledge cloud relay
Cross-device — sync memory across laptop, desktop, and server
Single binary — written in Rust, installs with one curl command
Privacy first — runs locally by default, cloud is optional and ephemeral
Context synthesis — Pro feature for intelligent context briefings

Supported AI Agents

Works with Claude Code, Cursor, Cline, Windsurf, Claude Desktop, Copilot CLI, Gemini CLI, OpenClaw, Roo Code, Continue, Codex CLI, Goose, Amp, Kiro, Trae, OpenCode, Factory, Antigravity, and Kilo Code.

Pricing

Free: 100 memories, local-only, no account required ($0). Standard: unlimited memories, 3 devices, cloud sync ($10/mo). Pro: unlimited everything, context synthesis ($20/mo).

Integrations

Documentation

Full documentation available at docs.ctxovrflw.dev.