Buildrack

Faster to capture than to forget.

Buildrack is a project shelf for AI builders. Track what you built, how you built it, and what it costs — from the terminal where you already work.

One command captures your stack, AI tools, connected services, and live URLs. One command shows your total monthly burn across every project. One command shares your build with a permanent public link.

Quick Start

npx buildrack init

That's it. Buildrack scans your project, detects the stack, and creates a project card on buildrack.vercel.app — no forms, no browser tabs, no context switching.

Why

You ship faster than you remember. After your 4th project this week:

• What stack did you use for that one?
• Which Supabase project belongs to which app?
• What services are still running — and costing you money?
• What's the live URL, the admin panel, the repo?

You could use Notion. You could make a spreadsheet. But you won't update either one mid-build.

Buildrack lives where you work. The CLI captures in seconds. The web reads and shares.

How It Works

$ cd ~/projects/my-app
$ buildrack init

  scanning project...

  name:      my-app
  stack:     Next.js, TypeScript, Tailwind CSS, Supabase
  ai tools:  Claude Code
  services:  Vercel, Supabase

  create this project on buildrack? (y/n): y

  → created: https://buildrack.vercel.app/dashboard/projects/abc123
  → project id saved to .buildrack

The detection engine reads config files — package.json, tsconfig.json, .env key names, directory markers — and infers your stack. It never reads source code, never reads .env values, never sends anything without your confirmation.

Commands

Capture

| Command | What it does | |---|---| | buildrack init | Scan project, create card | | buildrack push | Re-scan, show diff, update card | | buildrack scan | Dry run — show what would be detected, send nothing |

Quick-add

buildrack add service stripe 29       # → added service: stripe ($29.00/mo)
buildrack add service vercel           # → added service: vercel ($0/mo)
buildrack add tool "cursor"            # → added tool: cursor
buildrack add url live https://my.app  # → added url: live → https://my.app
buildrack add note "MVP launched"      # → note appended

No confirmation needed — these are explicit, atomic commands.

Overview

$ buildrack list

  my-app          active    $4.50/mo
  arpghub         active    $0/mo
  side-thing      paused    $2.00/mo  ⚠
  old-landing     archived  $7.00/mo  ⚠
  ──────────────────────────────────────
  total                     $13.50/mo
  dead spend                $9.00/mo  ⚠

The ⚠ flag marks paused or archived projects still costing you money. That's the whole point.

$ buildrack check

  my-app
  status:    active
  stack:     Next.js, TypeScript, Tailwind CSS, Supabase, Zod
  ai tools:  Claude Code, Cursor
  services:  Vercel ($0/mo), Stripe ($0/mo), Resend ($0/mo)
  cost:      $4.50/mo
  urls:      live  → https://my-app.com
             repo  → https://github.com/user/my-app
  public:    no
  notes:     MVP uses inquiry-based booking

Share

$ buildrack share
  → public url: https://buildrack.vercel.app/p/my-app
  → copied to clipboard

One command, one permanent link. Replaces posting the same launch description on five different platforms.

Other

| Command | What it does | |---|---| | buildrack login | Authenticate via browser (one-time) | | buildrack logout | Clear stored token | | buildrack open | Open project card in browser | | buildrack status | Alias for check |

Detection Engine

Buildrack reads config files to infer your stack. It is deterministic, transparent, and conservative.

What it reads:

| File | Detects | |---|---| | package.json | Dependencies → stack (Next.js, React, Vue, Svelte, Tailwind, Supabase, Stripe, Prisma, Drizzle, Zod, and 10 more) | | tsconfig.json | TypeScript | | .env / .env.local / .env.example | Key names → services (Stripe, Resend, AWS, OpenAI, Sentry, etc.) | | .vercel/ / vercel.json | Vercel | | supabase/ | Supabase | | .claude/ / CLAUDE.md | Claude Code | | .cursor/ | Cursor | | Cargo.toml / pyproject.toml / go.mod | Rust / Python / Go |

What it never reads:

• Source code files (.ts, .tsx, .js, .py)
• .env values — only key names
• Anything outside config files

Run buildrack scan to see exactly what the engine detects before sending anything.

Trust

• Open source — audit what it does
• Confirmation before send — init and push always ask y/n
• No secret values — .env key names only, never values
• No source code — config files only
• No telemetry — zero tracking, zero analytics
• No background processes — runs when you call it, exits
• .buildrack file — contains only the project UUID, safe to commit

Prerequisites

| Requirement | Version | |---|---| | Node.js | >= 18 |

No other dependencies. The CLI uses native Node.js APIs only.

CLI Reference

| Command | Description | |---|---| | buildrack init | Scan project and create a card | | buildrack push | Re-scan and update card with changes | | buildrack scan | Dry run — show detections, send nothing | | buildrack check | Print current project card | | buildrack status | Alias for check | | buildrack list | List all projects with total cost | | buildrack add service <name> [cost] | Add a connected service | | buildrack add tool <name> | Add an AI tool | | buildrack add url <label> <url> | Add a named URL | | buildrack add note <text> | Append to project notes | | buildrack share | Get public URL, copy to clipboard | | buildrack open | Open project card in browser | | buildrack login | Authenticate via browser | | buildrack logout | Clear stored token | | buildrack --help | Show help | | buildrack --version | Show version |

License

MIT