Your personal memory, across sessions, agents, and devices.
Your agents forget context between sessions. memU gives every agent you use one shared, inspectable memory layer with source-linked recall.
Fetch https://memu.so/SKILL.md by using API Key xxx as a Bearer token in the Authorization header. Then follow the retrieved instructions to install and configure memU, using the same key wherever required.
The durable constraints every agent should know before it starts.
Why the product story changed and what should stay consistent.
Patterns learned from code sessions and review history.
How each agent should retrieve and use personal memory.
Repeated mistakes that should not follow the user into another agent.
# Paste into a new Codex session
Fetch https://memu.so/SKILL.md by using API Key xxx as a Bearer token in the Authorization header. Then follow the retrieved instructions to install and configure memU, using the same key wherever required.
# Codex can now retrieve
# project preferences, launch decisions,
# coding habits, and tool patternsmemU carries preferences, decisions, and learned skills across the agents and workflows you use.
A convention learned in Codex stays available when the next task starts in Cursor, Claude, or another connected agent.
Carry preferences, decisions, and personal context across conversations, agents, and devices.
Keep research context and source-linked decisions available when work moves to another agent.
Reuse customer context and durable support playbooks without replaying every prior conversation.
Preserve goals, progress, and reusable guidance across sessions instead of starting over each time.
Keep voice, constraints, references, and creative decisions consistent between drafts and agents.
We work with design partners to customize memory, deploy it on the right infrastructure, and connect the agents their teams already use.
Shape memory scope, retrieval behavior, storage, and embedding choices around the way your teams and agents actually work.
Choose managed cloud or a supported self-hosted pilot with customer-controlled storage and embedding infrastructure.
Connect the agents your organization already approves through packaged host adapters or a scoped custom integration.
Prove the deployment and integration path with your real agents and data boundaries before expanding across the organization.
Enterprise capabilities are delivered through scoped design-partner engagements. Deployment, integration, and support terms are agreed for each pilot.
Release notes, design decisions, and practical thinking about memory across agents.
The harness has moved into the model. memU's July release deletes the pipelines, prompts, and LLM retrieval chains: the agent decides what to remember; memU just stores, embeds, and retrieves — in 474 lines, with zero LLM calls.
Claude Agent Skills package runbooks and guardrails as reusable files. Static skills still need MemU for memory of real outcomes across teams.
CI harnesses around Claude Code make agent quality measurable. Without MemU, harness scores do not compound into long-term debugging memory.
Meet the developers, contributors, and researchers building memory that can travel across agents.
Install memU once. Retrieve the right context in Codex, Claude, Gemini, OpenClaw, Hermes, and more.
Keep every memory inspectable and linked to its source.