AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
What happens between a cashier pressing Charge and money arriving — and why the callback cannot be trusted. An independent teardown by Genmars.
Developer toolsTypeScript · CSSGitHub @genmarstechOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
Loading the full README. The saved excerpt is shown below.
Frontend work Genmars builds — client projects, product prototypes and
design implementations. One folder per project, each standing on its own.
One folder per project, independent. Each has its own package.json,
its own lockfile and its own stack. cd into one, install, run.
That is deliberate rather than lazy. These are separate pieces of work for
separate briefs, and a workspace monorepo would tie a grocery dashboard's
React version to whateveRead the full README ↗Frontend work Genmars builds — client projects, product prototypes and design implementations. One folder per project, each standing on its own.
| Project | What it is | Stack |
|---|---|---|
mercato-admin | Grocery store operations dashboard — orders, inventory, delivery slots, team | Vite · React · TypeScript |
mercato-store | The same shop's storefront — aisles, basket, checkout, order tracking | Vite · React · TypeScript |
nero-26 | Formula 1 concept site — a car modelled in code, scroll-driven teardown, 3D tyre | Vite · React · TypeScript · three |
mvule-co | Nairobi furniture shopfront — KES pricing, WhatsApp ordering, M-Pesa deposits | Vite · React · TypeScript |
stk-lifecycle | R&D — interactive teardown of the M-Pesa STK push lifecycle, and why the callback cannot be trusted | Vite · React · TypeScript |
njia | Nairobi matatu arrival board — six real corridors, a deterministic morning, bunching detection, a map drawn with no map library | Vite · React · TypeScript |
shamba | Offline-first field survey — field-level CRDT merge on Lamport counters, and what whole-record last-write-wins quietly discards | Vite · React · TypeScript |
ratiba | Duty roster solver — backtracking with MRV in a cancellable Web Worker, and an explanation when no rota exists | Vite · React · TypeScript |
kioo | The Genmars design system — tokens, eight components, and contrast measured in the browser rather than documented | Vite · React · TypeScript |
mizani | Columnar query engine — dictionary-encoded store, a hand-written SQL subset, and a plan that says what it read | Vite · React · TypeScript |
deewee | DeeWee Natural Oils — a hair-oil shopfront built from a Claude Design handoff, with a two-question routine quiz and a draggable before/after | Vite · React · TypeScript |
One folder per project, independent. Each has its own package.json,
its own lockfile and its own stack. cd into one, install, run.
That is deliberate rather than lazy. These are separate pieces of work for
separate briefs, and a workspace monorepo would tie a grocery dashboard's
React version to whatever lands next. Coupling projects that need not agree
buys a smaller node_modules and costs the ability to pin anything.
Two of them share a brand rather than a build. mercato-admin and
mercato-store are the staff and customer halves of one shop and carry the
same palette and the same three typefaces — each copied into its own
theme.ts, not imported from a shared package. A duplicated palette is the
cheaper mistake: extracting it would make a colour change in one deploy a
rebuild of the other, for two projects that are each finished.
Each project carries its own README with what it is and how to run it.
They are demonstrations, and each one is built around a single engineering argument it can make better than a sentence can:
njia — that a city-scale map needs a projection, not a map library,
and that a simulation whose history depends on how you arrived at a
moment cannot be used to argue about anything.shamba — that the merge rule most sync layers ship converges
perfectly and is still losing data, and that the list of six discarded
observations is the argument the abstract one never wins.ratiba — that the valuable output of a solver is the one where it
fails, and that "no solution found" is true and useless.kioo — that a design system's accessibility figures should be
measured rather than written down. It found a real failure in its own
tokens the first time it ran.mizani — that storing down the columns is worth it, and that a
query plan which reports what it actually touched is how you know.Charter 03 §I runs through all of them. Between the five there is no map
library, no SQL parser, no dataframe library, no charting library, no CSV
library, no constraint solver and no CRDT library. The only runtime
dependency any of them has beyond React is three, in nero-26.
projects/<name>/
README.md what it is, how to run it, where the design came from
package.json with a `verify` script CI can call
Give it a verify script that does whatever "this is not broken" means for
that stack — typecheck and build, at minimum. It is the one convention
worth holding across otherwise independent projects.