Developer platform · open source core

More than a coding agent.

Six products on one open source core. Write the code, review it, run agents on it, and ship it, with one memory of your work behind all of them.

Prefer to inspect first? Inspect script · Fully open source

Code

The agent in your terminal that already knows your codebase, so you stop briefing it every morning.

memcode · terminal

↺ readymain · sonnet · auto · aurora · mirror · extra mile
● session
→
memcode · v0.36.0 · main · git 3✎ +120/-8 · ↑12.3k ↓4.1k · ctx 68% · sonnet · auto

Stop re-explaining your codebase

Other agents burn your first ten minutes and your tokens rediscovering what you told them yesterday.

See the plan before the diff

It shows you the approach, you approve it, and what lands matches what you approved.

Nothing new to install

One binary in the terminal you already have open. No IDE plugin, no browser tab to babysit.

Code Review

Every pull request reviewed before a human opens it, with findings your team will actually act on.

acme/payments · pull request #482
internal/broker/limit.go
89- if calls > b.maxCalls {
89+ if calls >= b.maxCalls {
HighCall cap now rejects the final allowed call

With a cap of 2 the second call is refused, so every build that uses its full budget fails at the last step.

Suggested fix
if calls > b.maxCalls {
Confirmed by CITestBrokerEnforcesCallCap
Nothing posted on the other 6 changed files.

Catch bugs before your team does

Every comment names the failure and the path to it, so the fix is obvious.

A bot nobody mutes

It stays silent unless something is actually wrong, so your team keeps reading it.

Your standards, not generic ones

Set what matters per repository and it holds that line on every pull request.

Agents

Put an agent where your team already talks, and give it work that gets done while you are somewhere else.

memcode admin · agents

◉ gateway3 agents · 2 channels · 3 schedules
→
memcode admin · gateway running · support ✓ · scout ✓ · ops ✓

Fix a bug from your phone

Message it on Slack, Telegram, or ten other channels. It does the work and replies in the thread.

Progress while you sleep

Hand over an objective and a schedule, then come back to work that is already done.

Nothing runs without your say

Unattended work happens only inside the permissions you granted up front.

Chat

Every model in one browser tab, with the same memory of your work that the terminal agent has.

New chat· 2 docs in context
payments-overview.md
What is our refund policy for EU customers on annual plans?

Here is what your docs say.

The policy

- Annual plans get a full refund inside the first 30 days, no questions asked

- After 30 days, refunds are prorated to the unused months

- EU customers are always refunded through the original payment provider

Worth knowing

- The 30-day window is a legal requirement in the EU, not just policy

- Prorated refunds need a support ticket today; self-serve is on the roadmap

Source

- payments-overview.md, refunds section, updated in April

2 docs

Every model in one place

Move between OpenAI, Anthropic, Google and the rest mid-conversation without losing the thread.

Your whole team, no terminal

The people who do not live in a command line get the same agent and the same context.

It knows your company

Answers come from your own documents, not just from what the model was trained on.

Datahub

The knowledge your team already has, in one place, so people ask the system instead of asking you.

Docs
NameSize
Architecture Decisions18 items
—
Runbooks6 items
—
Product Specs11 items
—
api-reference.md
28 KB
payments-overview.md
12 KB
deploy-runbook.md
15 KB
onboarding.md
6 KB
schema.sql
9 KB

Stop answering the same question

Upload the runbooks and decisions once, and let people get the answer without your time.

Answers you can check

Every reply cites the document it came from, so nobody has to take it on faith.

Context that outlives people

Architecture decisions and the reasoning behind them stay with the team when someone leaves.

Websites

Describe the site you want and get a real one live, without a designer, a build step, or a deploy.

ember-coffee.memcode.app
Ember Roasters
BeansStoryVisitOrder

Small-batch since 2019

Coffee roasted the slow way

Single-origin lots, roasted to order and shipped within 24 hours of the roast.

Order beansOur story

Ethiopia Guji

peach · bergamot

Huila Decaf

cocoa · plum

Ember Blend

toffee · citrus

A live site from one sentence

Say what you want and it ships. No template to fight, no framework to learn.

Change it by asking

Edit in plain language instead of hunting through files to find the thing you meant.

Nothing to host

It goes live on our infrastructure the moment you are happy with it.

Models

Every major model behind one agent, so you never rebuild your workflow when the leader changes.

OOpenAIAAnthropicGGoogleXxAIKKimiZGLMLOllama

Never locked to one vendor

OpenAI, Anthropic, Google, xAI, Kimi, GLM, or a local model. Switch without rebuilding anything.

Pay model cost, not markup

One prepaid balance across every vendor, debited at what the model actually costs. Or use your own keys.

The model you picked stays picked

Nothing swaps it out mid-task and bills you for a choice you did not make.

Open source.

The agent and the gateway are on GitHub under MIT. Self-host it, or let us run it.

One binary, MIT

The agent and gateway core are one Go module. Read it, fork it, build it.

Self-host it

Run the gateway yourself, or point the CLI at any compatible endpoint. No account needed.

No lock-in

Your keys, your models, your data on disk. Leave anytime.

Install in one line

macOS, Linux, and Windows. Runs in your terminal. Use Memcode credits, bring your own keys, or self-host the gateway.

Get started
MemcodePublic Beta