Exocortex

Every time you start a conversation with an AI agent, it knows nothing about you. Not your name, not your role, not what you're working on, not what you care about. You start from zero, every single time.

That means every conversation is generic. The agent can't challenge your assumptions because it doesn't know what you assume. It can't keep you focused because it doesn't know what you're focused on. It can't align its work with your values because it doesn't know what you value.

You are strangers, every time.

An exocortex fixes this. It's a structured external memory — a set of files that give your AI agent persistent context about who you are, what you're working on, and why it matters. Your agent reads these files at the start of every session and becomes a collaborator that knows you, challenges you, and holds you accountable.

You are not the author of your own thoughts. But you can write the prompt that prompts you back.

The vision

Every employee at Auki has an exocortex. Every function has an exocortex. Every project has an exocortex.

Individual exocortices carry identity (including values), role, goals, and attention. They make each person more productive and more aligned with the mission.

The organization's exocortex — shared files like organization.md, skills/, and team role files — gives every agent on the team the same foundational context through a symlinked org repo. Alignment isn't a quarterly exercise. It's continuous, maintained by the system itself.

When we get this right, our intercognitive bandwidth increases: our ability to think, experience, and solve problems together with each other and AI.

Three levels

The exocortex exists at three levels, each a different scope of shared context:

• User level — a single person and their AI agent. Where identity, role, goals, and daily attention live. Each person has their own personal exocortex repo.
• Organization level — context shared across everyone at the company. Mission, strategy, team, conventions. Lives in the org repo (e.g. aukilabs/org) and is symlinked into each person's exocortex as org/.
• Project level — an individual repo or piece of work, usually with its own team. Lives in that project's own repo (e.g. aukilabs/relay).

When working, you open your personal exocortex and symlink in the organization repo plus whichever project directories are active for you. Your agent then reads across all three levels to build its context.

Structural files

Each level carries the same three structural files:

• AGENTS.md — the entry point for the AI agent at that level. Session behavior, read order, alignment rules.
• CONTRIBUTING.md — logging conventions (changelog). May extend what the enclosing level says.
• CHANGELOG.md — append-only record of what changed and why, latest on top.

And one identity file per level, answering "what is this?":

Level	Identity file
User	user.md + user_role.md
Organization	organization.md
Project	vision.md (distinct from GitHub's README.md)

vision.md at the project level is separate from README.md because a project repo's root README.md is GitHub's landing page — that slot is already doing discovery and setup work. vision.md holds the aspirational "why this exists" without competing for it.

Current state

The src/ directory contains the exocortex template — the set of files that any Auki employee can clone to set up their own exocortex. See src/README.md for a file-by-file description and how setup.sh works.

See roadmap.md for the project roadmap and sprint.md for the current week's focus.

Setup

You’ll need

• A GitHub account.
  • Auki employees: access to the private aukilabs/org repo.
  • Public: clone the public aukilabs/org-auki repo — it is the public counterpart to org (same role in your exocortex; symlinked as org/ after setup).

Steps

1. Clone repositories Clone this exocortex repo and one org repo next to it (same parent directory). Open your terminal and go to the desired parent directory, for example aukilabs (create it if needed).

 cd ~/aukilabs

Auki employees:

 git clone git@github.com:aukilabs/org.git
 git clone git@github.com:aukilabs/exocortex.git

External contributors:

 git clone git@github.com:aukilabs/org-auki.git
 git clone git@github.com:aukilabs/exocortex.git

2. Run the setup script In your terminal, go into the exocortex project. Then run the setup script.

 cd exocortex
 bash src/setup.sh

The script is interactive. It will ask where to create your personal exocortex (don't choose the same aukilabs directory), then ask simple questions: your name, role, what you care about, what you’re working toward, and what you’re focused on today. Your responses will be recorded in the starter files for you. 3. Open your new exocortex in Cursor or give it to your agent When the script finishes, navigate to where you specified your personal exocortex should be and open that folder in Cursor. Over time you’ll edit the markdown files there; the AI reads them at the start of sessions so it stays aligned with you. Alternatively, you can give your AI agent access to your exocortex and let it manage it for you. 4. Link your active projects Your exocortex is also where your active project work lives. After setup, open org/src/projects.md inside the org repo you cloned — the canonical list of project repos where applicable. Clone the projects relevant to your work into a sibling directory, then symlink each one into your exocortex root:

 ln -s ~/path/to/project-repo ~/my-exocortex/project-name

Every project follows a standard structure (readme.md, roadmap.md, src/sprint.md, etc.) so the agent always knows where to look. To create a new project from scratch, see org/src/CONTRIBUTING.md § Creating a new project. 5. Keep shared context fresh When the org repo is updated, pull the latest changes in the clone you symlinked (your exocortex always sees it as org/). In the terminal, go to that directory — e.g. org or org-auki next to exocortex — and run git pull.