Sheetify Studio

Sheetify Studio is a multi-tenant SaaS platform that lets teams upload, build, and share Streamlit applications without exposing a terminal. It combines a FastAPI control plane, Docker-based runners, and a modern Next.js dashboard so creators can iterate quickly while operators retain control.

Stack

• Backend – FastAPI + SQLAlchemy (async) with Celery workers and Redis for job orchestration.
• Runner – Docker sandbox orchestrated by a dedicated FastAPI service that builds tool images from a hardened Python 3.11 + Streamlit base and advertises routes through Traefik under /t/<tool_id>.
• Frontend – Next.js (TypeScript) with Tailwind CSS for the dashboard, build logs, and lifecycle controls.
• Data stores – PostgreSQL for persistent state, Redis for queues.

Features

1. Email/password authentication with JWT sessions.
2. CRUD for Streamlit tools (name, description, owner) and version uploads via .py or .zip artifacts.
3. Packaging pipeline that normalises uploads into app.py + requirements.txt, infers dependencies when missing, and blocks dangerous imports such as subprocess, os.system, and socket.
4. Build jobs executed via Celery that produce container images from the trusted base, install requirements, and persist build logs + image references.
5. Runtime orchestration to start/stop Streamlit containers with streamlit run app.py --server.headless true --server.baseUrlPath /t/<tool_id>; Traefik publishes each tool under /t/<tool_id>.
6. Guardrails including import filtering and a base image entrypoint that drops outbound network traffic by default.
7. Next.js dashboard with a “Create Tool” wizard (paste or upload), build log viewer, start/stop controls, and shareable URLs.
8. A ready-made CSV Explorer template in templates/csv_explorer/app.py to seed new deployments.

Getting started

1. Build the sandbox base image (required for the runner to layer user code):

   docker build -f runner/base/Dockerfile -t sheetify-base:latest .

2. Launch the stack:

   docker compose up --build

   This brings up PostgreSQL, Redis, the FastAPI API, Celery worker, runner service, Next.js dashboard, and Traefik proxy.
3. Access the dashboard at http://localhost. The backend API is proxied under /api by Traefik.

API surface

Method	Endpoint	Description
POST	/v1/auth/register	Create a new user.
POST	/v1/auth/token	Obtain a JWT access token.
POST	/v1/tools	Create a tool.
GET	/v1/tools/{id}	Fetch tool details, versions, builds, and runs.
POST	/v1/tools/{id}/versions	Upload a new version (.py or .zip).
POST	/v1/tools/{id}/build	Queue a build for a version.
POST	/v1/tools/{id}/run	Start the latest build in the sandbox.
POST	/v1/tools/{id}/stop	Stop and remove the running container.

Frontend flows

• Create Tool – authenticate, complete the wizard (paste code or upload archive), and you’ll be redirected to the tool detail view.
• Build & Run – from the tool page you can queue builds, watch logs update in near real-time, start/stop the Streamlit container, and copy the share link exposed at /t/<tool_id>.

Development notes

• The Celery worker and runner communicate over HTTP; update RUNNER_URL in docker-compose.yml if you change hostnames.
• Runner containers inherit security controls (no-new-privileges, drop NET_RAW, outbound firewall) from the hardened base image.
• Tool uploads are scanned for banned imports before storage.
• Extend the template catalog by dropping additional apps into templates/.

Testing

Run the backend unit test suite with pytest from the repository root:

pytest

If you add JavaScript/TypeScript tests in the frontend, execute them with the appropriate npm script from the frontend/ directory (for example, npm test).

Happy building!