AI Fabrix - Builder

Open developer runtime for AI Fabrix ecosystem. Install the AI Fabrix platform and test it locally. Then add external integrations or build your own applications.

← Full documentation: docs/README.md (table of contents for all guides)

Why AI Fabrix Builder?

• Build perspective: Everything is driven by declarative config and JSON schemas—no hidden logic, AI assistant–friendly.
• Industry standards and security: Follow industry standards and high security (e.g. ISO 27k); no secrets in version control.
• Full lifecycle in your version control: Configuration, apps, and integrations live in your own VCS (GitHub, GitLab, Azure DevOps).
• One tool from day one: Single CLI for local infra, app and integration creation, build, run, and deploy—same workflow for apps and integrations.
• Consistency and production readiness: Schema-driven; deploy apps and integrations to the same controller/dataplane; production-ready secrets with kv:// and Azure Key Vault.
• Application development: Use miso-client (TypeScript and Python) to talk to the dataplane and controller. The repo includes both TypeScript and Python SDKs; see templates/applications/dataplane/README.md and the repo for usage.

Prerequisites

• Node.js 18+ – Recommended for running the CLI.
• AI Fabrix Azure / platform: Install from Azure Marketplace or run via Docker (e.g. aifabrix up-platform). You need full access to Docker (docker commands) where applicable.
• Secrets before platform: Add secrets (e.g. OpenAI or Azure OpenAI) before running aifabrix up-platform; the platform reads them from the place you configure. See Infrastructure and secrets configuration.

Install

npm install -g @aifabrix/builder

Alias: You can use af instead of aifabrix in any command.

Goal 1: Start and test the AI Fabrix platform

Get the platform running locally so you can try it.

One-shot install:

aifabrix setup

aifabrix setup detects your local state and either runs a fresh-install wizard (admin email/password, optional AI tool keys) or shows a mode menu (re-install, wipe data, clean files, update images). It then runs up-infra and up-platform for you. Use aifabrix teardown to fully remove the local installation. See Infrastructure commands for details and CI flags.

If you prefer to set the AI tool key outside the wizard, use one of:

• OpenAI: set your API key:

  aifabrix secret set secrets-openaiApiKeyVault <your-openai-secret-key>

• Azure OpenAI: set endpoint and API key:

  aifabrix secret set azure-openaiapi-urlKeyVault <your-azure-openai-endpoint-url>
  aifabrix secret set secrets-azureOpenaiApiKeyVault <your-azure-openai-secret-key>

Secrets are stored in ~/.aifabrix/secrets.local.yaml or the file from aifabrix-secrets in your config (e.g. builder/secrets.local.yaml).

→ Infrastructure guide

Goal 2: External system integration

Create and deploy an external system (e.g. HubSpot): wizard or manual setup, then validate and deploy.

Example: HubSpot

• Create: aifabrix create hubspot-test (external is the default; or aifabrix wizard for guided setup). For a web app use aifabrix create my-app --type webapp.
• Configure auth and datasources under integration/hubspot-test/.
• Validate: aifabrix validate hubspot-test
• Deploy: aifabrix deploy hubspot-test

→ External systems guide · Wizard

Goal 3: Build your own application

Create, configure, and run your own AI Fabrix application locally or deploy it (create app → configure → build → run / deploy).

→ Your own applications

Documentation

All guides and references are listed in docs/README.md (table of contents).

• CLI Commands Reference – All commands
• Infrastructure – What runs and why
• Configuration reference – Config files (deployment key, application.yaml, env.template, secrets)

Requirements

• Docker Desktop – For running containers
• Node.js 18+ – For running the CLI
• Azure CLI – For deploying to Azure (optional)

License

© eSystems Nordic Ltd 2025 - All Rights Reserved

@aifabrix/builder is part of the AI Fabrix platform.