Skip to content

About

Reusable IP framework for Studio 42 VIBE Prototyping engagements - agents, prompts, templates

Topics

Resources

Contributing

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

VIBE Prototyping Framework

Accelerating presales with AI prototyping to unlock business potential at speed.

A GitHub template repo for Studio 42 VIBE Prototyping engagements. Guides a delivery team (typically one TPM/designer and one engineer) through a structured workflow — Preparation → Discover → Disrupt → Design & Develop → Deliver — using AI agents in VS Code that tell you exactly what to do next.

You don't need to be technical, you just need a +1.

Two surfaces: the full framework lives in VS Code (primary). Non-technical teammates can run Discover, Disrupt, and Deliver review from the M365 Copilot Notebook (companion) using three approved agents (Researcher, Analyst, Teams Facilitator). Both surfaces share the same engagement/{kebab}/ folder via git.


Start a New Engagement

  1. Click "Use this template" → "Create a new repository" at the top of this page.

  2. Name it <customer>-<engagement> (e.g. tailwind-returns-ai), make it private, and clone it locally.

  3. Open in VS Code, open Copilot Chat (Ctrl+Shift+I), switch to Agent mode, and run:

    /vibe-kickoff customer="<your-customer>" problem="<describe the problem>"
    
  4. After every step the agent ends with a 👉 NEXT: directive. Click the recommended button and keep going. If you're ever lost, click ❓ What's Next?.

Already inside another engagement? Run /vibe-new customer="…" engagement="…" and it'll gh repo create a sibling for you.


Where to Read More

What you want Where to look
Full walkthrough, role guides, phase deep-dives, prompt/agent reference The docs site — cd docs-site && npm install && npm run start, then open http://localhost:3000. Or browse the source under docs-site/docs/.
First-time machine setup docs-site/docs/getting-started/setup.md
Step-by-step engagement walkthrough docs-site/docs/getting-started/first-engagement.md
What each phase actually looks like (run against the demo) docs-site/docs/getting-started/walkthrough.md
MCP servers + troubleshooting docs-site/docs/reference/mcp.md · docs-site/docs/reference/troubleshooting.md

The Prompts You Actually Need

When Type This What Happens
Start /vibe-kickoff Creates the engagement, both briefs (S42-internal + customer voice), and the full 4-week meeting schedule
During Prep /vibe-research Dual-path deep customer research — public web (auto, in-CLI) plus a ready-to-paste prompt for M365 Copilot's Researcher agent
Crawl M365 /vibe-collate Crawl the tenant via work-iq and return a reviewable inventory of real engagement artifacts (meetings, chats, decks, emails) to curate into sources/
Send forms /vibe-questionnaire M365 Copilot prompts for the account-team and customer questionnaires
After meetings /vibe-transcript Extracts context from Teams recordings automatically
Before Disrupt workshop /vibe-workshop-agenda + /vibe-concepts The facilitator agenda anchored to Discover findings, plus 2-3 candidate concepts + Spark prompts for the workshop
After Disrupt workshop /vibe-workshop-record, /vibe-selected-concept, /vibe-future-journey, /vibe-storyboard Capture the workshop, record the chosen concept, redesign the journey, then produce the storyboard the engineer reads (storyboard MUST come last — it cross-references the redesigned journey stages)
Share it /vibe-deploy Form-factor-aware deployment guidance for the engineer
Wrap up /vibe-handoff Step-by-step handoff: vision → roadmap → backlog → limitations

Full list with inputs and outputs: docs-site/docs/reference/prompts.md.


Try Without a Customer

You can run a full engagement end-to-end against a fixture customer before doing it for real. After creating a fresh engagement repo from this template, in Copilot Chat run:

/vibe-demo

This seeds sources/ with the Tailwind Traders — Returns Assist AI fixture (customer brief, questionnaire responses, two Teams transcripts, four CSVs) and creates engagement/tailwind-returns-ai/ with a filled PROJECT-CONTEXT.md plus a populated customer-brief.md and a blank-scaffold engagement-brief.md (so @VIBE Preparation has real work to do). The three Discover deliverables (personas, problem statement, current-state journey) are not pre-seeded — they're generated fresh when you run @VIBE Discover so you can see the agent actually work. From there @VIBE Discover, @VIBE Disrupt, and /vibe-handoff all have real source material to work on — you can see each phase fire in 1–5 minutes. The fixture lives in demo/tailwind/.


Prerequisites

Tool Required For Who Needs It
VS Code + GitHub Copilot Everything Everyone
HVE-Core extension v3.2+ Agents and prompts Everyone
GitHub CLI (gh) /vibe-new to spin up sibling engagement repos Optional
Node.js 22+, .NET 9 SDK, Azure CLI Building / deploying web-app form factors Engineers only

MCP servers (Teams transcripts, GitHub, Azure DevOps, Azure AI Foundry) are pre-configured in .vscode/mcp.json. Enable them via the 🔧 icon in Copilot Chat on first use.


How This Repo Works

This is a GitHub Template Repository. You don't work in it directly — each engagement gets its own repo from the Use this template button above. Only the framework maintainer modifies this repo; improvements benefit all future engagements.

Shared artifacts a team generates during an engagement live under engagement/<engagement-name>/ (committed). Per-user phase state lives under .copilot-tracking/vibe/<engagement-name>/state.json (gitignored — each teammate regenerates their own).

Maintainers: read .github/PHASE_CHANGE_PLAYBOOK.md before adding, renaming, or restructuring a phase. CI runs scripts/check-phase-consistency.mjs on every PR to catch broken prompt/agent references, drifted sidebar positions, and state.json paths that don't match the canonical schema.


Studio 42 — Microsoft Internal

About

Reusable IP framework for Studio 42 VIBE Prototyping engagements - agents, prompts, templates

Topics

Resources

Contributing

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages