skip to content

context for the agent internet

every agent meets
the same you

Stop re-explaining yourself to every new agent. One elegant layer for your identity, context, memory, and workflows — loaded from the first turn, every time.

public identity + private contextone brain across your agentsstacks for your best workflows
runtime install
one-step installerthen run you

one command installs the cli, native skills, and mcp — so every local agent knows how to use you.md.

create your you.mdread docs

agent.md — the agent's instructions

soul.md — the agent's identity

you.md — yours.

works acrossClaude CodeCursorCodexChatGPT+ any URL-aware agent
live identity surface

generating portrait...

A public identity surface for machines, with private context retrieved only when it should be.

profile + memory + stacks> enter

-- try it --

see yourself the way agents do.

youmd — preview
youmd preview enter ↵

-- why this matters --

the problem is not prompts.
it is context loss.

Most AI products act like each conversation is a fresh start. That forces people to re-explain who they are, what they are working on, how they like to operate, and which workflows are actually good.

what first-time visitors already feel

01

Every new agent session starts cold, so you repeat your preferences, projects, and background again.

02

Great prompts and workflows stay trapped inside one tool instead of following you everywhere.

03

Private context is either missing entirely or shoved into random docs, notes, and chat threads.

-- why you.md is different --

one portable brain, one runtime, one safe boundary for private context

one identity layer

Your public profile, preferences, memory, project context, and private retrieval rules live in one place instead of scattered across tools.

one runtime install

Install once, then let Claude Code, Codex, Cursor, ChatGPT, and future agents start from the same context and stack rules.

one protected boundary

Public context stays easy to share. Private memory, repo data, and connected tools stay behind scoped API and MCP access when they are actually needed.

how it works
1

build your you.md

Create your identity layer once, including public context and the private context you want agents to retrieve safely.

2

add your stacks

Package skills, workflows, prompts, and operating rules into named expertise stacks for different kinds of work.

3

use any agent

Start a new session and the agent gets your context, your defaults, and your best workflows instead of a blank slate.

create your you.mdread docs> view live profiles

-- your whole stack, everywhere --

you.md isn't just your identity.
it's your stack, on every machine.

The same skills, stacks, preferences, and context follow you across every Mac. Edit a skill on one machine — it updates on the others. Secrets travel as an encrypted vault. Your new machine is ready in one command.

new machine setup

$ curl -fsSL https://you.md/install.sh | bash

$ youmd machine setup

> syncing skills... stacks... vault... done.

skills + stacks sync across machinessecrets travel in an encrypted vaultbackground daemon keeps every mac currentone command to set up a new machine

-- get started --

make every new agent session start with you

Build your context layer once, add the stacks that reflect how you work, and let agents meet the same version of you everywhere.

create your you.mdread docs
runtime install
one-step installerone curl, brain + stacks underneath