Documentation & analysis center

Read it, test it, question it.

Protocols, provenance rules and limitations; an AI research analyst that stays separate from measured data; a verification suite you can run right here; and the live project maturity panel.

Documentation library

AI research analyst

AI-generated · not measured data

Explains concepts, summarises run configurations, compares recorded datasets and points out missing controls. Every reply is labelled as interpretation. Calls run server-side; no API keys are in this page.

Include the latest recorded runs from the registry as context

How the analyst is grounded

It receives the project description, the integrity rules and — optionally — up to 30 recent registry records with their provenance.
It is instructed to separate measured (records), simulated (browser reference) and interpretive statements, and to refuse to invent results, citations or hardware data.
The repository configures a Gemini-powered analyst behind a serverless endpoint. On this deployment the assistant uses the hosting platform's built-in model through a server-side call — the same separation: keys never reach the browser.

⚠

Treat replies as hypotheses. AI output can be wrong. It never upgrades a result's status; only verified evidence changes the maturity panel.

Project maturity

Status, by evidence.

Loading…

Labels move only when verified evidence becomes available. The project owner maintains them in the MaturityItem table.

Verification suite

Automated checks of the critical mathematical operations used by the browser reference implementation. They run in your browser in under a few seconds.

Not run yet.

Configuration & deployment

The source project is a React + TypeScript application with Vercel serverless endpoints. Typical setup — verify names against the repository README:

git clone <repository>
npm install
npm run dev          # local development
npm test             # unit tests for math & circuit passes
vercel --prod        # deploy (env vars set in Vercel project)

Server-side environment (never in browser code)

Purpose	Typical variable	Needed for
AI analyst	GEMINI_API_KEY	Gemini analysis endpoint
AWS credentials	AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY	Braket submit & poll
AWS region	AWS_REGION	Device availability
Results bucket	BRAKET_S3_BUCKET	Task output storage
Target device	BRAKET_DEVICE_ARN	Simulator or QPU

If a variable is missing, the corresponding endpoint should return a clear “not configured” error — and this site labels the integration as not configured rather than simulating it.