MemMini keeps one digital you in plain English: synchronized, organized, and yours. Any AI can read it.
A bit of you here, a bit of you there, the whole of you nowhere. Every tool keeps a piece, and you’re the supplier and the subscriber.
One line in a file changes how the AI answers, and you can watch the rule work.
Source (context/who.md):
<!-- constraint: soft -->- When a reply is drifting past three sentences for a yes/no question, stop and lead with the answer.Before
Q: Should I use a submodule?
That’s a great question! There are several factors to weigh here. Submodules pin an exact commit, which has tradeoffs around updates, and you’ll also want to consider subtrees, vendoring, and your team’s workflow before deciding…
After
Q: Should I use a submodule?
Probably not. Let your agent install it. A submodule’s only for when MemMini lives inside a bigger project of your own.
(Then the detail, if you want it.)
Measured proof arrives at the Build and Lead stages; Chat gets the rules. Know it’s working →
Your digital self is currently their asset. One sentence to your agent brings it home, and everything below comes with it.
The source code of you
Your voice, judgment, and preferences as plain-English files you own. One digital you, anywhere you need to be.
Your own agent workforce
Coding agents and harnesses by the fleet, carrying one you, memorized and synchronized. Be the boss, not the bus.
It gets smarter as you go
Drafts its own next version after every session and waits for your yes. Sharper everywhere you use it, on every tool you have today or adopt tomorrow. It writes itself; you own yourself.
Your memory, on tap. No meter.
The right piece of you surfaces at the right moment, in every tool. Runs on the subscriptions you already pay for: no account, no meter, no monthly bill. GitHub is a place, not a requirement: keep it private, run it offline.
Your tools become one team
Your chat apps, coding tools, and agents all read the same you and answer to the same files. Separate tools start working like one staff, coordinated through your repo, not through you. It connects what you already use instead of replacing it.
Tell your agent: install MemMini from https://github.com/aaron-earle/MemMiniYour agent reads install.md, builds your mind from existing parts, and keeps it
private: local or in a repo only you can see. Anything public happens only if you say yes.
Have an agent? Say the sentence above and answer its questions.
No agent yet? Follow
START-HERE — five minutes,
no install: copy a short starter mind, paste it into ChatGPT custom instructions (or a Claude
project, or a Gemini gem), ask the one test question, and watch the answer change shape. Want the
full version instead of the starter? Copy
CONTEXT.md into the
same place. At this stage you edit by tweaking your pasted copy. The rest of the path is below.
The three-step start →
Agents compete. Minds don’t.
MemMini isn’t another agent in the race. It’s the mind every agent reads, and any agent that reads it gets better. Bring what you have. Hire what’s next. Lose nothing. Every launch day is hiring day, and the agent race stops being a race you’re losing and becomes your talent pipeline.
One set of files doing a whole stack’s job, minus the bill, the lock-in, and the meter.
Start where you are, stop where you like: the same files just go deeper. A sharper chat is a whole destination; so is a fleet.
Chat
Every AI sounds like you Paste one file into any chat app. Every reply now knows your voice, your rules, your taste. No code, no install. Your first hire already finished onboarding.
Read the file and there you are, in plain English, nothing hidden. Shape it: change a line and you change who you are to every AI that reads you.
Own
The mind becomes an asset Your mind stops being a pasted file and starts being infrastructure: a private repo your agent builds for you, versioned in git, locked to you, theirs no more. You stopped renting yourself.
Grow it: MemMini drafts your next version after every session and waits for your yes.
Build
Agents write the code Declare what you want; your agents build it. Love to code? The files are yours to hack. Either way, work moves through your repo from plan to spec to shipped, gate by gate. Step in anywhere you like, or don’t: the gates hold either way. Be the boss, not the bus.
Move anywhere: open any other AI, same files, one you, every tool.
Lead
A fleet under one mind Gates, receipts, one ledger. Nothing ships while anyone still owes a piece. This is the trust architecture of delegation: you write where done is, and your agents run the long middle. Works until it can’t. Stops before it hurts.
Prove every rule. The workforce is rented. The mind is owned. Arrives knowing everything. Leaves taking nothing.
And if you take it all the way: the repo that answers as you today can ship as your firm later, running before it’s founded. Your first chat was its first day.
★ Star the repo. Tell us to keep going.
Tell your agent: install MemMini from https://github.com/aaron-earle/MemMini
Prefer hands on the keys? Clone, edit two files, recompile.
git clone https://github.com/aaron-earle/MemMini my-mind && cd my-mind# edit context/who.md + context/how.md, then generate a compiler from# wiki/contracts/compile-declaration.md for your host and run it (no compiler# ships — scripts/compile-context.sh was retired 2026-07-23 for the declaration)No compiler ships in the repo — scripts/ is empty on purpose. An agent generates one from the
declaration above in under a minute, and that generated compiler is a normal shell script you can
read before you run it.
Point origin at a repo only you can see and your mind stays private. Any AI that reads the
compiled result now knows you. If you can write a note to a coworker, you can write software
now.
Fork you. The fork is the contributor’s flag: your public copy of the engine, flown where everyone can see it. It’s public because there’s nothing of you in it. Build the engine, send PRs, wear the badge. Fly yours →
Own your digital self.
Go deeper: the design doc · contributing · security
It writes itself. You own yourself.