Skip to content

yatarousan0227/vibe-todo

Repository files navigation

VibeToDo

日本語版 README

This repository is using speckit-for-projects https://github.com/yatarousan0227/speckit-for-projects

Turn vague project ideas into actionable plans — with AI.

VibeToDo is a local-first web app that takes your rough project concepts or daily work descriptions and transforms them into structured, executable plans. You capture the intent; the AI handles the refinement.


Why VibeToDo?

Most planning tools assume you already know what you want to build. VibeToDo starts where you actually are — with a fuzzy idea — and guides you through structured intake, AI-powered refinement, and task synthesis.

Your vague idea  →  Structured intake  →  AI refinement  →  Actionable tasks

Intake screen


Features

  • Dual planning modes — Switch between project and daily_work to match what you're planning
  • Hybrid input — Combine structured fields (title, objective, scope) with free-form context in a single draft
  • Persistent drafts — Save anytime and resume later with a projectId URL parameter
  • Review before confirming — Inspect your intake in review state and edit before locking it in
  • Refinement-ready handoff — Confirming intake initializes a workspace context for downstream AI refinement
  • Pluggable LLM — Supports OpenAI, Anthropic, and Azure OpenAI via a unified provider adapter

Tech Stack

Layer Technology
Framework Next.js 16 + React 19
Language TypeScript
Database PostgreSQL
Testing Vitest
Infrastructure Docker / Docker Compose

Quick Start

1. Install dependencies

npm ci

2. Configure environment

cp .env.example .env

Edit .env to set your database URL and LLM provider:

# Database
DATABASE_URL=postgres://vibetodo:vibetodo@localhost:5432/vibetodo

# LLM Provider — choose one: openai | anthropic | azure_openai
LLM_PROVIDER=anthropic
ANTHROPIC_API_KEY=sk-ant-...
ANTHROPIC_MODEL=claude-sonnet-4-6

3. Start PostgreSQL

docker compose up -d db

4. Initialize the schema

npm run db:init

5. Start the dev server

npm run dev

Open http://localhost:3000.


Docker (all-in-one)

Start the app and database together:

docker compose up --build
Service Port
App 3000
PostgreSQL 5432

Testing

npm test

Uses vitest run. Integration and E2E helpers are also available:

npm run test:integration
npm run test:e2e

API Reference

POST /api/projects

Save a draft or confirm intake.

generationTrigger Behavior
"draft_save" (or omitted) Saves as draft
"intake_confirm" Confirms intake and initializes refinement session

Example — save a draft:

{
  "generationTrigger": "draft_save",
  "project": {
    "planning_mode": "project",
    "structuredInput": {
      "title": "New onboarding flow",
      "objective": "Reduce setup friction",
      "background_or_current_situation": "Users drop before first success",
      "scope_summary": "Focus on first-run experience",
      "stakeholders": "PM, Design, Support",
      "expected_outcome_or_deliverable": "",
      "constraints_or_conditions": "Keep rollout low-risk"
    },
    "freeFormInput": {
      "body": "Free-form planning context goes here."
    }
  }
}

GET /api/projects/:projectId/workspace-context

Returns the workspace context for a saved draft or confirmed project. The right panel of the UI renders this response directly.


UI Behavior Notes

  • Draft ID is not assigned until first save
  • Append ?projectId=<id> to auto-resume a session
  • Review state is self-contained — no page navigation required
  • allowedActions.canConfirm reflects current state

Directory Structure

app/
  api/projects/          Next.js Route Handlers
  page.tsx               Home screen
src/
  components/
    intake-app.tsx       Intake UI
  lib/intake/            Domain logic, validation, persistence
scripts/
  init-db.ts             DB schema initialization
briefs/                  Requirement briefs
designs/                 Design artifacts

Roadmap

Phase Feature Status
SCR-001 Project intake & draft management ✅ Implemented
SCR-002 AI-powered spec refinement workbench ✅ Implemented
SCR-003 Task plan synthesis ✅ Implemented
SCR-004 Management workspace (kanban / timeline) ✅ Implemented

SCR-002 — Spec Refinement Workbench Draft every artifact, edit directly in place, and move through approval without leaving the screen.

Refinement workbench

SCR-004 — Management Workspace Kanban and Gantt views driven from a single source of truth, linked back to your approved artifacts.

Kanban view


Caveats

  • Authentication is not implemented — this is a local MVP
  • Designed for general project and work descriptions, not software-specific only

Documentation


Contributing

See CONTRIBUTING.md.

License

MIT


VibeToDo is an MVP. Rough edges are expected and intentional.

About

VibeToDo is a local-first web app that takes your rough project concepts or daily work descriptions and transforms them into structured, executable plans. You capture the intent; the AI handles the refinement.

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors