Today the first open-source vibe-coding platform, tomorrow the next geneation "IDE" for Web Applications.
announcement_01_gh.mp4
Make sure to check the Stripe Integration section before running these commands. As of now, you need to provide all the Stripe keys in order to have the commands running smoothly.
# run postgres and minio
cp .env.example .env
docker compose up -d
# run the database migrations
npm run db:push
# seed the db with a User.
npm run db:seed
This will create the following user:
- User:
admin@example.com
- Password:
admin12345
You can, of course, create new users as well through /sign-up
.
For the storage of static files such as profile pictures we are using MinIO. For this reason, after you have the docker-compose up and running, you can visit http://localhost:9001
with the .env
credentials and create anew bucket uploads
which will be used by the web application and make it public.
Finally, run the Next.js development server:
npm run dev
For project previews, we use Docker containers to isolate and run Next.js applications. You need to make sure to have docker installed and pull the right image:
docker pull ghcr.io/filopedraz/kosuke-template:v0.0.5
This image will be used for all project previews, providing consistent development environments for your generated applications.
For handling subscriptions, we use Stripe. Add these environment variables to your .env
file:
STRIPE_SECRET_KEY=sk_test_...
STRIPE_WEBHOOK_SECRET=whsec_...
NEXT_PUBLIC_STRIPE_PUBLISHABLE_KEY=pk_test_...
NEXT_PUBLIC_APP_URL=http://localhost:3000
To test webhooks locally, you can use the Stripe CLI to forward events to your local server:
stripe listen --forward-to localhost:3000/api/webhooks/stripe
This will provide you with a webhook secret that you can use in your .env
file.
To set up the linting pre-commit hook:
npm install
npm run prepare
This configures a Git pre-commit hook that runs linting and prevents commits with issues. To bypass in exceptional cases:
git commit -m "Your message" --no-verify
- Brand Guidelines Generator: include in the pipeline a step related to Font and color palette generation based on the project description.
- Ollama Support: support Ollama as backend for code generation.
- Database and backend support: build full-stack web applications, not only static pages.
Kosuke uses a sophisticated agentic pipeline for code generation that operates through a series of structured phases:
-
Initialization & Context Gathering
- The system processes the user request and initializes the agent with a project ID
- Project context (directory structure) is automatically gathered
- Recent chat history is retrieved to provide continuity
-
Thinking Phase (Iterative)
- The agent enters an iterative "thinking" mode to gather information
- It can read files to understand the codebase structure and context
- The system carefully tracks which files have been read to avoid redundant operations
- Context accumulates with each iteration, building a comprehensive understanding
- Safeguards prevent infinite loops by limiting iterations (max 25)
-
Execution Phase
- Once sufficient context is gathered, the agent switches to "execution" mode
- A structured list of actions is generated (create/edit/delete files or directories)
- Each action is executed sequentially using appropriate tools
- After execution, a concise summary of changes is generated and presented
Kosuke is licensed under the MIT License.
For questions or support, you can create an issue in the repo or drop me a message at filippo.pedrazzini (at) joandko.io