Builder OS

Name: builder-os
Author: v60samurai

An operating system for shipping products solo with AI.

PRD → Brand Guide → Session Playbook. The three documents I use to take a product from idea to live URL, working alone with Claude Code.

These templates have shipped real products. They are opinionated. They tell you what to do, not what you could do.

What's inside

Folder	What it is	Use when
`prd/`	Two PRD templates: lean (15KB) and full (48KB)	You're choosing what to build
`brand/`	Two brand-guide templates: quick and full	You're choosing how it looks
`sessions/`	A 14-session implementation playbook	You're building it
`examples/`	Real filled-out examples from shipped products	You want to see one done

Which template, when

Cohort submission in 2 weeks ─────→ prd/lean-prd.md + brand/quick-brand-guide.md
Shipping in 1-3 months ───────────→ prd/full-prd.md + brand/full-brand-guide.md
Already have a PRD, want to build → start at sessions/SESSION_PLAYBOOK.md
Just need design rules ───────────→ brand/ standalone
Need a worked example to copy ───→ examples/

Quickstart

Copy the templates into your project:

npx degit v60samurai/builder-os my-product
cd my-product

Then:

Open prd/lean-prd.md (or full-prd.md for serious projects). Search for {{ and fill every placeholder. Use the confidence tags (🟢🟡🔵🔴) honestly.
Open brand/quick-brand-guide.md. Decide your colors, type, voice. Write down why you chose each one.
Open sessions/SESSION_PLAYBOOK.md. Follow the sessions in order. Do not skip checkpoints.

A solo builder using this end-to-end ships in roughly 12-16 hours of focused work.

Philosophy

Most templates are slop.

They give you sections without telling you what a good answer looks like. They use placeholder text that an AI happily fills with confident nonsense. They optimise for the writer feeling done, not for the reader making decisions.

Builder OS does the opposite:

Every section tells you why it exists, what good looks like, and what bad looks like.
Confidence tags (🟢 primary research, 🟡 secondary, 🔵 hypothesis, 🔴 disproven) force you to own how solid each claim is.
Brand decisions get defended in writing. If you can't write why, the choice isn't made yet.
The session playbook has checkpoints. Discovering a broken foundation in session 8 costs 3x what it costs in session 2.

The result: when you hand these docs to Claude Code (or a teammate), they make the decisions you would have made. No defaults. No drift. No slop.

What this is NOT

A framework. There's nothing to install.
A SaaS. No login, no pricing, no roadmap dictated by a Stripe dashboard.
A starter kit. No code. Bring your own stack.
Beginner-friendly. Assumes you've shipped at least once and know what a PRD is for.
Stack-agnostic. The session playbook assumes Next.js + Supabase + FastAPI. Adapt as needed.

Roadmap

Templates I'll add next (PRs welcome):

discovery/ — pre-PRD research synthesis (interview notes → insights)
voice/ — brand voice corpus + microcopy patterns
adr/ — architecture decision records
postmortem/ — shipping retro template
eval/ — MVP completeness rubric
growth/ — post-launch experiment template
.claude/ — starter CLAUDE.md, commands, hooks for Claude Code
.cursorrules — equivalent guard for Cursor users

See CHANGELOG.md for what's already shipped.

Contributing

Issues, PRs, and new template ideas are welcome. Read CONTRIBUTING.md first. The voice is opinionated, and PRs that dilute it won't merge.

License

MIT. Harshit Badiger (@v60samurai).

Use these in any project. Fork them. Sell them. Rip them apart. If they ship something for you, a star (and a tag on the post you ship from them) is the thanks I'd love.