assessment-model/README.md

This is a Next.js project bootstrapped with create-next-app.

Getting Started

When first getting set up you'll firstly want to install the existing dependencies. To do this, simply run

npm install
# or
yarn install

First, run the development server:

npm run dev
# or
yarn dev
# or
pnpm dev

Open http://localhost:3000 with your browser to see the result.

You can start editing the page by modifying app/page.tsx. The page auto-updates as you edit the file.

This project uses next/font to automatically optimize and load Inter, a custom Google Font.

Learn More

To learn more about Next.js, take a look at the following resources:

• Next.js Documentation - learn about Next.js features and API.
• Learn Next.js - an interactive Next.js tutorial.

You can check out the Next.js GitHub repository - your feedback and contributions are welcome!

Deploy on Vercel

The easiest way to deploy your Next.js app is to use the Vercel Platform from the creators of Next.js.

Check out our Next.js deployment documentation for more details.

We currently have a development version, found at https://assessment-model-dev.vercel.app, and a production version at https://assessment-model.vercel.app, however the production version is missing a significant number of environmental variables, which will need to be added.

Drizzle ORM

We're using Drizzle ORM to interface with our AWS Postgres database. Documentation on getting set up can be found here

Schemas

In order to get started with Drizzle, a schema needs to be created. Schemas can be added src/app/db/schema as typescript files. See the documentation on how to set up schemas but effectively, Drizzle allows you to define schemas as typescript code, which allows for simple, type safe schema definitions.

Creating Migrations

To create a migration, a command has been set up in package.json. Simply run

npm run migration:generate

Or with yarn/pnmp accordingly.

Note, there seems to be a bug with Drizzle which is documented here.

The workaround is to open up tsconfig.json and comment out "target": "es5", and replace it with "target": "ESNext". This should hopefully only be a temporary workaround required.

Pushing migrations

To push migrations, another command has been set up in package.json, since drizzle-kit currently does not support pushing for Postgres out of the box.

Run

npm run migration:push

Which will commit changes to the database. The database changes will be pushed to the public schema, whereas a meta record will be pushed to the __drizzle_migrations schema.

Inserting users into the database

In order to insert a user into the database, simply run

npm run create_users -- {email} {firstName}

Since we're using just the built in process arguments to read command line arguments, the ordering of arguments needs to be email address and then name

Cypress Testing Documentation

This document provides an overview of how to perform end-to-end testing using Cypress for the login functionality of the application. The testing code is based on the provided code snippets.

Prerequisites

Node.js installed on your machine Cypress installed as a dev dependency in your project

Test Execution

To execute the login tests, follow these steps:

1. Run

npm run test:e2e:open

Which will open Crypress test runner

2. Select the tests that you want to run. At the time of writing, only login tests have been completed

Key files

The key files that are at play for testing are documented here. Because of some issues testing next-auth and setting cookies, a custom command to set the JWT, and avoid the functionality defined in the signIn function as defined in src/app/api/auth/[...nextauth]/route.ts

cypress/plugins/index.js is a standard file, required by cypress to journey through the Google Oauth flow

cypress/figtures/session.json is a user fixture that is used to log in a user in the login tests

cypress/support/commands.ts creates a custom login user command which sets a JWT and allows us to actually authenticate. cy.intercept only mocks the client side behaviour of the apis and therefore does not set any cookies, do this function does this manually.

impeccable live mode

Live mode lets you pick an element in the running app, drop a comment, and have variants swapped in via HMR. Start it from Claude Code with:

/impeccable live

Then open the app on the forwarded port 3000 and use the picker.

The skill is vendored at .claude/skills/impeccable/ rather than installed per-container, so live mode's scripts/ are on a stable path and everyone gets the same version. Refresh it with npx impeccable update.

Two things are easy to get wrong in the devcontainer:

• Port 8400 must reach the container on host port 8400 exactly. The script injected into the page hard-codes http://localhost:8400 (only the port is configurable, not the host), so if it lands anywhere else the picker just never appears — no error, no mark on the page. Two paths, both covered: docker-compose.yml publishes 8400:8400 (this is what devcontainer.sh needs — the devcontainer CLI does not implement forwardPorts, that's a VS Code-only feature), and devcontainer.json pins it with requireLocalPort for the VS Code path. Because the mapping is fixed, only one container at a time can run live mode.
• Live mode edits src/app/layout.tsx to inject its script tag, and strips it again on exit. If a session is killed uncleanly, run node .claude/skills/impeccable/scripts/live-server.mjs stop to remove the tag — don't commit it.

DESIGN.md and PRODUCT.md are the context live mode generates variants against (DESIGN.md wins on visual decisions, PRODUCT.md on strategic/voice ones), so keep them honest.

Playwright

Playwright sits alongside Cypress rather than replacing it. Specs live in e2e/ (Cypress keeps cypress/e2e/), and playwright.config.ts boots next dev for you unless something is already serving http://localhost:3000.

npm run test:playwright

The devcontainer image bakes in chromium and its OS libraries, so there is nothing to install on a fresh container. Chromium is the only browser baked in — adding firefox or webkit to the config means adding them to .devcontainer/Dockerfile too. The @playwright/test version in package.json and PLAYWRIGHT_VERSION in that Dockerfile are pinned to the same number on purpose: Playwright ties each browser build to a library version, so if they drift, runs fail with "Executable doesn't exist".

Watching the browser

Runs are headless by default. To watch one drive, start the virtual desktop (Xvfb -> fluxbox -> x11vnc -> noVNC, mirroring the Model devcontainer), open the forwarded port 6080 in your browser, then run headed against that display:

bash scripts/start_viewer.sh
DISPLAY=:99 npm run test:playwright:headed
bash scripts/start_viewer.sh stop

Generating pre-signed urls

In our terraform stack, we have a module called s3_presignable_bucket which contains the definition for our bucket which we will use to store retrofit plan input csv's in.

We will generate a pre-signed url and then make a post request to that endpoint to store that data to s3. Part of that process is the creation of an AWS IAM role which contains the permission set to access the bucket, rerofit-plan-inputs-<stage>. The name of this IAM role is s3_presign_role_<stage> and for our NextJS application, as it's hosted outside of AWS (for the moment), we need to generate a set of access credentials to give the application access to this bucket. The access key and secret key are automatically generated and stored in AWS secrets manager under dev/presign_frontend/access_key and dev/presign_frontend/secret_key and need to be set in the environment for the pre-sign api to store csv data to aws.