KnowledgeMay 13, 2026·1 min read

Wax — On-Device Memory Layer + MCP Server

Single-file memory layer for AI agents (Swift) plus an MCP server for local-first RAG. Good for private, on-device workflows on Apple Silicon.

AI Open Source
AI Open Source · Community
Agent ready

This asset can be read and installed directly by agents

TokRepo exposes a universal CLI command, install contract, metadata JSON, adapter-aware plan, and raw content links so agents can judge fit, risk, and next actions.

Needs Confirmation · 62/100Policy: confirm
Agent surface
Any MCP/CLI agent
Kind
Memory
Install
Npx|Spm
Trust
Trust: Established
Entrypoint
npx -y waxmcp@latest mcp install --scope user
Universal CLI install command
npx tokrepo install 44829892-4fe7-561a-9356-6067d0d30410
install contractmetadata JSONadapter planraw content
Intro

Wax is a local-first memory layer for agents: keep retrieval on-device and expose it via an MCP server so tools can write/read memory without prompt scaffolding.

Best for: Apple Silicon users who want private, low-latency memory for agent workflows

Works with: MCP clients (Claude Code, etc.); SwiftPM integration (see README)

Setup time: 5–20 minutes

Key facts (verified)

  • README includes an MCP install command via npx -y waxmcp@latest ....
  • Also documents HTTP transport flags for remote/team setups (see README).
  • GitHub: 731 stars · 42 forks; pushed 2026-05-13 (GitHub API verified).

Main

Treat memory as a real interface

Write small, queryable facts, keep writes explicit, and log reads/writes in your agent traces.

README excerpt (verbatim)

Wax Banner

Wax is a high-performance, single-file memory layer for AI agents on Apple platforms.
On-device, private, and portable. No server and no cloud dependency.

Swift Platforms License Stars

English · Español · Français · 日本語 · 한국어 · Português · 中文

What is Wax?

Wax is a Swift-native persistence engine for AI agents. It stores documents, embeddings, and structured knowledge in a single portable .wax file.

The goal is simple: keep memory local, keep setup light, and make recall fast enough that it can stay in the loop.

Why Wax?

Feature Wax SQLite (FTS5) Cloud Vector DBs
Search Hybrid (Text + Vector) Text Only* Vector Only*
Latency ~6ms (p95) ~10ms (p95) 150ms - 500ms+
Privacy 100% Local 100% Local Cloud-hosted
Setup Zero Config Low Complex (API Keys)

FAQ

Q: Is Wax an MCP server or a library? A: Both—README documents an MCP server install path and SwiftPM integration.

Q: Do I need a separate vector database? A: The README positions it as local-first; treat it as an on-device retrieval layer for many cases.

Q: Can I host it remotely? A: README shows HTTP transport options; evaluate privacy and auth before using remote mode.

🙏

Source & Thanks

Source: https://github.com/christopherkarani/Wax > License: Apache-2.0 > GitHub stars: 731 · forks: 42

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets

MemoryGraph — Graph MCP Memory Server (pipx)

MemoryGraph is a graph-based MCP memory server for coding agents; verified 199★ and shows a pipx install + `claude mcp add` setup for Claude Code.

Knowledge
Agent Toolkit

Wax — Single-File Memory Layer for AI Agents

Wax stores documents, embeddings, and knowledge in one portable `.wax` file, giving AI agents a local memory layer without extra servers.

Knowledge
AI Open Source

Recall — Persistent Memory MCP for Claude Agents

Recall provides persistent cross-session memory for Claude agents via an MCP server backed by Redis/Valkey or managed cloud, with sharing and search.

Knowledge
MCP Hub

yantrikdb-server — Memory DB with Contradiction Checks

YantrikDB adds persistent agent memory with contradiction checks. Use `yantrikdb-mcp` for Claude Code/Cursor, or scale with a server cluster.

Knowledge
AI Open Source
◈Home🔍Search👤Me