Get started

Brunel documentation

Everything Brunel does, precisely. From a GitHub repository to a priced, validated production infrastructure, with the REST API, the MCP server and the enterprise controls.

Brunel is an AI infrastructure architect. You paste a GitHub repository; Brunel reads the code, tells you what it runs with the file that proves each fact, designs its production infrastructure in four tiers priced from the cloud providers' official catalogs, and generates validated Terraform, CI workflows and documentation.

Start here

  • Quickstart: analyze a repository and get four priced architectures in about two minutes.
  • Core concepts: projects, facts, context, tiers, versions and the rules that keep designs honest.
  • Plans and limits: what each plan includes, generated from the plan catalog.

What Brunel does

StepWhat happensRead more
AnalyzeThe repository is downloaded read-only and 16 deterministic detectors extract facts, each with its evidenceFacts and evidence
CheckKnown vulnerable dependencies, committed secrets, Dockerfile, CI and existing IaC risksSecurity check
DesignFour complete tiers covering eight domains, mapped to real AWS or Scaleway servicesArchitecture tiers
PriceUnit prices from the official catalogs, every line shown with its catalog dateHow prices are computed
RefineAsk in plain language; every change is a new version with a diff and a cost deltaRefine
GenerateTerraform per domain, Kubernetes or Compose, GitHub Actions, runbook and ADRs, all validatedGenerated files
OperateRe-analysis on every push, pull request reviews, price watch, deployment through your CIReal-time tracking
ProveSigned reliability certificates anyone can verifyReliability certificates

Build on Brunel

  • REST API: the same actions as the app over HTTPS, authenticated with an API key.
  • MCP server: let Claude Code, Cursor, VS Code, Windsurf or any MCP client use Brunel as a tool.
  • Webhooks: signed events for tracking alerts, pull request reviews, deployments and more.

Principles you can rely on

  • Evidence first. Every detected fact links to the file and line that proves it, and you can correct any of them.
  • Deterministic where it matters. Detection, mapping, pricing, validation and file generation are code, not AI output.
  • Your CI, your credentials. Deployment runs in your GitHub Actions; Brunel never stores cloud keys.
  • Untrusted input stays data. Repository content is wrapped as data for the AI, which has no tools and cannot follow instructions found in code.

Where things live

AddressWhat it is
https://brunel.cloudThe product: chat, analyses, designs
https://console.brunel.cloudPlatform: API keys, MCP, usage, logs, webhooks, certificates
https://enterprise.brunel.cloudEnterprise: members, security and SSO, audit log, billing
https://portal.brunel.cloudPrograms for startups, education, open source and partners
https://api.brunel.cloudREST API (/v1) and MCP server (/mcp)
https://docs.brunel.cloudThis documentation
Brunel documentation: from repository to production infrastructure