Wax
Sub-Millisecond RAG on Apple Silicon. No Server. No API. One File. Pure Swift
claude mcp add --transport stdio christopherkarani-wax npx -y waxmcp@latest mcp install --scope user
How to use
Wax is an on-device memory management solution for Apple platforms that stores AI agent memories in a single portable .wax file without relying on servers or cloud services. The MCP workflow allows you to install and manage Wax as an MCP server so you can interact with its memory capabilities from your tooling. Once installed, you can initialize a memory store, remember user preferences, and perform recall operations to retrieve context-aware results, all while keeping data on-device for privacy and offline usage. The MCP setup exposes an installable Wax server via npm (waxmcp) which you run with npx to integrate Wax into your project’s MCP ecosystem. Typical usage involves opening or creating a memory store, storing memories, and recalling relevant context for prompts, with Wax handling hybrid retrieval, on-device embeddings, token budgeting, and crash-safe persistence all in a single file.
How to install
Prerequisites:
- Node.js and npm installed on your development machine
- Access to the project or environment where the MCP server will be registered
Step-by-step installation:
-
Ensure Node.js and npm are installed. You can verify with:
- node -v
- npm -v
-
Install the Wax MCP server via npx (no global install required): npx -y waxmcp@latest mcp install --scope user
-
Confirm installation and load Wax as an MCP server within your MCP tooling. The command in step 2 registers Wax under the scope you specified (e.g., user) so you can reference it from your MCP workflows.
-
(Optional) If your environment requires a local cache or specific Wax file location, configure the path to the .wax file within your MCP client setup.
-
Integrate Wax usage into your application scripting, for example by creating a memory store, remembering items, and recalling relevant context as demonstrated in the Wax README.
Additional notes
Notes and tips:
- Wax stores memories in a single portable .wax file, enabling easy backup and transfer while keeping data on-device.
- The MCP installation command uses waxmcp via npx to avoid a global install; you can specify --scope to control access or project scope.
- For Claude or other AI integration, Wax provides clear memory-handling patterns (remember, recall, forget, context, reflect, handoff) that you can map to your MCP workflows.
- Ensure your app has adequate storage permissions and handles memory file creation in a writable directory (e.g., app sandbox on iOS/macOS).
- If you encounter compatibility issues, verify that your Node.js/npm versions meet the requirements of waxmcp and that the Wax package dependencies resolve correctly within your environment.
Related MCP Servers
ai-trader
Backtrader-powered backtesting framework for algorithmic trading, featuring 20+ strategies, multi-market support, CLI tools, and an integrated MCP server for professional traders.
mem0
✨ mem0 MCP Server: A memory system using mem0 for AI applications with model context protocl (MCP) integration. Enables long-term memory for AI agents as a drop-in MCP server.
shellguard
MCP server that gives LLM agents read-only shell access over SSH
limps
limps your Local Intelligent MCP Planning Server across AI assistants. No subscriptions, no cloud—run it locally. Version control your planning docs in git. No more context drift—one shared source of truth across Claude, Cursor, Copilot, and any MCP tool.
ultrasync
MCP server from darvid/ultrasync
backlog
Help coding agents and developers to keep track of a project's backlog by storing tasks as markdown in git.