memU skill
Long-term memory for agents: it distills knowledge from session history and brings it back in new tasks, one store shared across Claude Code, Codex, Cursor and
- 14,618 ★ on the repository
- Last commit
Extended access
Why this level
- Reads the agent's session history and edits the host instruction file
- Registers a scheduled background task and in cloud mode sends derived memory to MemU Cloud
Install
- Works in
Manual install
pip install --upgrade memu-cliPuts memu and the host adapter binaries on PATH. Then follow SKILL.md to configure.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reads the agent's session history and edits the host instruction file
- Registers a scheduled background task and in cloud mode sends derived memory to MemU Cloud
This is third-party code. Review the repository files before installing.
What it does
memU gives an agent shared memory across sessions, hosts and devices. It installs beside the agent as a set of memu-cli adapters: a scheduled background task reads the session history, and the agent itself turns it into short notes and reusable Markdown skills. Before answering, the agent runs retrieve and pulls in what is relevant to the current task, and a standing instruction in the host file (CLAUDE.md, AGENTS.md and similar) triggers this automatically. Memory lives locally in SQLite or Postgres, or in MemU Cloud, and one configured backend is shared by every connected agent. Setup is driven by the bundled SKILL.md: it detects which host you are, installs the package, prints the built-in guide and walks it step by step with verify gates.
Who it is for. For developers who work in coding agents daily and want context and learnings to persist across sessions and tools.
Good fit when
- You want the agent to remember project decisions and conventions across sessions
- You use several agents and want shared memory between them
- You want good moves from history to become reusable skills
Not a fit when
- You need a one-off task without installing background processes or patching the agent's instruction file
- Security policy forbids reading session history or sending memory to an external service
Example request
Read memU's SKILL.md and set up memory for this agent, keep the backend localLimitations
memU installs a scheduled background task, reads session history files and edits the host instruction file, and in cloud mode sends derived memory to MemU Cloud. Local mode needs an embedding provider and its key: openai by default, with jina, voyage, doubao and openrouter as alternatives. The support matrix varies by agent and operating system, and some modes are marked as working with limitations. Concurrent access and large stores need Postgres with pgvector.
How to disable. Ask the agent to read SKILL.md and follow the uninstall flow, or run <your-binary> docs uninstall. It removes the scheduled task and the instruction block; the memory store and ~/.memu/config.env are kept by default, and memory is erased only on explicit request.
FAQ
Is a cloud account required?
No. There is a local mode with memory in SQLite or Postgres on your device. MemU Cloud with a key from memu.so is a separate option.
Does memory work across different agents?
Yes, when they share the backend from ~/.memu/config.env. What one host's sessions taught memU, another host later retrieves.
Discussion
No questions or reviews yet.
Related
SkillsShared local memory for AI agents: Claude Code, Codex, Cursor and others find the agreements of past sessions in it and record new ones
Persistent memory for coding agents: insights are stored as markdown cards and recalled at the start of the next session.
A long-term memory plugin that keeps decisions, preferences and task state across sessions and injects only what a query needs
An MCP server that gives an agent memory across sessions, projects and machines: decisions, dead ends, preferences and skills
One local binary that gives Claude Code, Codex and Cursor shared memory across sessions, with a knowledge graph on disk and no API keys
A local workbench for Claude Code, Codex, Pi and OpenCode: memory changes wait for your review, then an audit follows each turn
README in short
The README describes memU as a lightweight agent memory system with a store shared across sessions, hosts and devices. It works through adapters for Codex, Claude Code, Cursor, OpenClaw, Hermes, WorkBuddy, Cola and pi, with a memu-agent detect command for other agents. Each adapter binds two seams: recording history into memory and injecting memory through the host instruction file. The package installs via pip, npx or uvx; memory lives in SQLite, Postgres or MemU Cloud; and the chosen provider computes embeddings. Apache-2.0 licensed, with the core memory logic in roughly 500 lines.
README badge
Original SKILL.md
---
name: install-memu
description: Install or uninstall memU for whatever agent you are — identify your host, print its packaged guide, and follow it to wire (or unwire) both seams (record and inject). Use when the user asks to install, set up, integrate, remove, or uninstall memU.
---
# Install memU
> **Audience: the agent.** A user has pointed you at this file ("read SKILL.md
> and follow it to install memU"). This skill does not contain the install
> steps — it routes you to the right guide, which ships inside the memU package
> and is always in sync with the installed code. Do not install from memory or
> from blog posts; print the guide and follow it to the letter.
memU integrates with a host agent through two seams: **record** (a scheduled
bridging task mines your session log into durable memory) and **inject** (a
standing instruction in your instruction file makes you retrieve before
answering). Each supported host has its own adapter binary carrying its own
install guide. Your job here is three steps: install the package, pick your
binary, print and follow its guide.
## Step 1 — install the package
```
pip install --upgrade memu-cli
```