Skip to article
Source of Truthfor Life Sciences / Developer guide
Explore the application ↗
Start / Developer documentation

Deploy the example

Run the complete local application with the PostgreSQL clone recipe. Its isolated app, database and explicit initializer give each clone its own synthetic records and expiring Source workspaces. Database choices explains the supported persistence boundary; fixture provenance explains the example data.

Choose a route

Route Current capability Operator responsibility
Local PostgreSQL containers Pinned app/database images, private project volume, explicit schema/seed validation and runtime role Docker/Compose and Node 22 for the clone CLI; follow local operations
Native Node with PostgreSQL Full static assets, GraphQL and stored histories/Source workspaces Provision database/roles, run explicit setup, inject runtime settings, manage process and backups
Node with database configuration absent Scoped GraphQL over bundled synthetic records and browser-session edits Build and start the Node server
Local Vite development Hot reload and bundled interface examples Install the frozen dependency graph; Vite does not host the Node APIs
Static browser assets Browser presentation without the Node APIs or durable workspaces Choose that limited scope explicitly, or provide a separately configured API
Standalone documentation build:foundation emits admitted articles at the root of dist-foundation/ Static publication, domain configuration and a receiving-host check

Ordinary Node hosting

Use the declared Node 22.x and pnpm versions. From the reviewed source:

pnpm install --frozen-lockfile
npm run build
PORT=3000 npm start

The server binds the process to the host interface on PORT. Keep dist/, build/server/, server/shared source and required runtime packages together. The full dependency graph also contains build tools; the Dockerfile separates build preparation from the non-root runtime image. Publishing dist/ alone omits /graphql and /workspace/graphql. vite preview is an asset preview rather than the complete application server.

For persistence, follow PostgreSQL setup or the native route in local operations. An ordinary database-owning setup login initializes the catalog and workspace schema. A distinct runtime login receives bounded data privileges. Normal application startup reads existing records; schema changes remain explicit operator work. The complete absence of database settings selects bundled examples. Partial settings and configured storage failures retain their visible failure state.

A remote operator owns HTTPS termination, process lifetime, network restrictions, database TLS and backups. The local clone recipe binds the published app port to loopback. The native Node server listens on its host interface, so use a firewall or reverse proxy for the intended audience. Local serving checks establish HTTP, source-file exclusion, finite GraphQL operations and startup behavior; they leave provider proxy, certificate and backup behavior for that provider's check.

Review site metadata before publishing a clone. Set the canonical url and social image for the intended origin. The build enables indexing and publishes selected documentation at /docs/, /llms.txt and /llms-full.txt; review those sources as public content. After deployment, observe one built asset, the mobile Corridor, Plate C3 and any installed workspace. Record the deployed source revision beside that result.

Source development and Figma Make

npm run dev and npm run dev:source run Vite with hot reload. The source retains .figma/make, .mise.toml and the Vite adapter for asset resolution, metadata and preview configuration. A fresh Make session owns its install/preview/import verification. Its existing static deployment script does not establish a Node GraphQL process. The root package and frozen lockfile remain the dependency owners.

Recover and continue

Preserve the previous reviewed source or image for application rollback. A source rollback leaves database records intact. The clone runbook supplies bounded stop/start, trusted backup restore and confirmed reset; another host supplies its own recovery process. Keep credentials, database dumps and browser cookies with their private owners.

Application delivery, schema setup, repository visibility and documentation-site publication are distinct actions. Copied workflows do not copy account settings, secrets, DNS or protection policies. Automation names current checks; a successful test supplies evidence for its selected behavior and does not publish the application.

Documentation · Repository

Repository references

These paths identify source in the clone. They are not application routes.

  • .figma/make
  • .figma/make/site.json
  • .mise.toml
  • Dockerfile
  • README.md
  • docs/development-automation.md
  • vite.figma.config.ts