Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Save egyjs/e2f43ca49f67af36e6f3b5d9a8dd34a1 to your computer and use it in GitHub Desktop.
Save egyjs/e2f43ca49f67af36e6f3b5d9a8dd34a1 to your computer and use it in GitHub Desktop.
Turn any repo into an AI-ready workspace in seconds. | StartVibe is a lightweight CLI tool that instantly sets up your project for AI-assisted, documentation-driven, and behavior-driven development.
title description keywords author
StartVibe: Bootstrap AI‑Ready Repos with Documentation‑Driven and Behavior‑Driven Development
Turn any repository into an AI‑readable workspace in seconds. Start curious, shape the idea, and ship a CLI that scaffolds /.ai docs, Copilot instructions, and BDD‑ready workflows.
StartVibe
start-vibe
AI-driven development
documentation-driven development
behavior-driven development
BDD
repository template
CLI
npx
GitHub Copilot
AI agents
egyjs

StartVibe — Give Your Codebase a Vibe

Start curious. Let your thoughts wander until a spark hits — that "what if?" moment. What if a repo could explain itself to an AI the way we explain ideas to teammates? What if documentation wasn’t a chore but the engine? Shape it into a concept, refine it into a workflow, and then build the tool that brings it to life.

That tool is StartVibe.

A tiny CLI that turns any repository into an AI‑ready, docs‑first, BDD‑friendly workspace — with one command.


TL;DR

  • StartVibe installs an AI‑native scaffolding into your repo: /.ai docs, Copilot instructions, and feature specs.
  • You write intent in Markdown. Agents (and humans) turn it into tests and code.
  • Works with any language; encourages BDD with your native test framework.
  • One command to get started:
npx start-vibe

Table of Contents


Why this, why now?

Modern codebases aren’t just for humans anymore — they’re read by AI copilots, CI bots, and automation. Yet most repos still bury intent in issues or chats. Developers guess; AIs hallucinate. Slow velocity, uneven quality.

StartVibe flips the default: make intent explicit, structured, and close to the code. Give both humans and machines a shared, living source of truth.

Benefits you feel immediately:

  • Faster onboarding (for humans and AI tools)
  • Clearer changes because the “why” lives alongside the “what”
  • Smoother BDD/TDD loops — specs become tests, tests shape code
  • Fewer miscommunications, fewer surprises

The concept: Vibe Coding

Vibe Coding is a simple idea:

  1. Describe behavior in plain language.
  2. Let agents (and humans) translate that intent into tests.
  3. Make the code satisfy those tests.

Documentation‑Driven Development (DDD) + Behavior‑Driven Development (BDD) + AI‑assisted iteration. The “vibe” is your repo’s personality — the guardrails that help agents act like great collaborators.


Core principles

  • Documentation as source of truth — Markdown first, code follows.
  • Behavior over implementation — focus on “When X, expect Y.”
  • AI‑readable by design — consistent structure, clear context, minimal ambiguity.
  • Framework‑agnostic — use your language’s native test tools.
  • Non‑destructive — lives in /.ai; your app code stays your app code.

The /.ai workspace (source of truth)

StartVibe installs a dedicated, discoverable knowledge space. It’s small, opinionated, and practical:

.ai/
  ARCHITECTURE.md          # System design, components, data flow
  CONVENTIONS.md           # Coding patterns, naming, style
  TICKETS.md               # Priorities, backlog, task tracker
  ci_cd_instructions.md    # How to build, test, and deploy
  features/
    example_feature_login.md  # BDD-style feature spec(s)
.github/
  copilot-instructions.md  # IDE assistant guidance

Why a separate folder? Two reasons:

  • Keep intent separated from implementation — clean mental model.
  • Give AI tools a single, predictable entry point.

From docs → tests → code

The loop looks like this:

  1. Author feature specs in /.ai/features/*.md using Given/When/Then.
  2. Generate native tests (Jest, Pest, RSpec, etc.).
  3. Implement code until tests pass. Repeat.

This respects your stack while making the intent crisp and machine‑consumable.


The CLI: start-vibe

StartVibe is a zero‑config bootstrapper. It drops the AI‑ready docs into your current repo without touching your application code.

  • Install via NPX (no global install required):
npx start-vibe
  • Key flags:
    • --force — overwrite existing files
    • --dry-run — show what would be created
    • --silent — suppress logs
    • --verbose — detailed progress

The CLI copies the templates and prompts before overwriting (defaults to Yes with a batch mode when desired). It’s fast, friendly, and gets out of your way.


Templates overview

You’ll get thoughtfully structured starter files, each with a purpose:

  • /.ai/ARCHITECTURE.md — system design and component boundaries
  • /.ai/CONVENTIONS.md — coding patterns and style
  • /.ai/TICKETS.md — task tracking and priorities
  • /.ai/ci_cd_instructions.md — build, test, deploy guidance
  • /.ai/features/*.md — behavior specs (Given/When/Then)
  • /.github/copilot-instructions.md — tuned hints for IDE assistants

These are starting points, not rules. Customize them to your team’s voice and needs.


Language and BDD framework mapping

StartVibe avoids lock‑in. Write specs in Markdown; generate tests in your ecosystem:

  • PHP → Pest
  • JavaScript/TypeScript → Jest
  • Python → Behave / pytest‑bdd
  • Java → Cucumber‑JVM
  • Ruby → RSpec
  • C# → SpecFlow
  • Go → Ginkgo
  • Swift → Quick

Generated tests belong in your project — not under /.ai — so they run in your normal CI.


Example feature flow

  1. Create a feature spec:
# 2025-11-01_feature_user_login.md

Feature: User Login
  As a registered user
  I want to log in securely
  So that I can access my account

Scenario: Valid credentials
  Given a registered user with email "[email protected]"
  When they submit valid credentials
  Then they are redirected to their dashboard
  And a session cookie is set
  1. Generate tests in your stack (manually or via an agent).
  2. Implement the feature until tests pass.
  3. Commit: docs, tests, code — the trifecta.

Roadmap and extensions

  • Language/framework detection to auto‑suggest test libraries
  • Generator that translates /.ai/features/*.md into native tests
  • GitHub Action to validate features on PR
  • Richer agent recipes in /.ai/agents.md
  • Optional /.ai/history/ to capture decisions and iterations

FAQ

Is this only for AI? No — humans benefit first: clearer intent, faster onboarding, better commit narratives.

Will it rewrite my code? No. StartVibe only adds documentation and guidance files. Tests and implementation remain in your app folders.

Can I use it in any repo? Yes. It’s framework‑agnostic and non‑destructive.

Is this BDD‑only? It encourages BDD because it pairs well with AI assistance, but you can adapt the flow to your style.


What I built

I packaged the idea into a real tool you can use today:

npx start-vibe

Give your codebase a vibe — and let AI understand it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment