Full‑Stack AI Agent Template

What it is – A command‑line generator that scaffolds a complete production‑ready web‑app (FastAPI backend + Next.js frontend) pre‑wired with AI‑agent capabilities, Retrieval‑Augmented Generation, authentication, billing, observability and deployment tooling. The generated project can be run locally with Docker or deployed to Kubernetes, and it ships a rich UI for chat, team/workspace management, knowledge‑base ingestion, and admin dashboards.

---

🎯 Core purpose

The template lets developers spin up a full‑stack AI‑agent SaaS in minutes, then focus on the domain‑specific logic instead of wiring together the usual plumbing (auth, DB, RAG, CI, monitoring, etc.).

---

🚀 Quick start (from the README)

1. Install the generator

   pip install fastapi-fullstack   # or `uv tool install fastapi-fullstack`
   

2. Create a project – answer the interactive wizard (or pass flags):

   fastapi-fullstack               # scaffolds `my_ai_app`
   

3. Bootstrap the stack

   cd my_ai_app
   make bootstrap                 # builds Docker images, starts DB, runs migrations, seeds admin
   

4. Run the UI (in another terminal)

   cd frontend && bun install && bun dev   # or `make dev-frontend`
   

5. Open the services:
   • API docs: http://localhost:8000/docs
   • Admin UI: http://localhost:8000/admin (admin@example.com / admin123)
   • Frontend: http://localhost:3000

---

🛠️ Main features (as listed)

Category	What you get
AI Agent frameworks	Support for 5 runtimes – Pydantic AI, Pydantic Deep Agents, LangChain, LangGraph, DeepAgents.
RAG pipeline	Plug‑and‑play vector stores (Milvus, Qdrant, pgvector, ChromaDB) with automatic chunking and embedding.
Web stack	FastAPI 15 backend with WebSocket streaming, Next.js 15 frontend, real‑time chat UI.
Conversation sharing	Public links, admin‑only browser view, export/share of chat sessions.
Enterprise‑ready	JWT + OAuth, admin panel, Celery task queue, Docker & Kubernetes configs, environment‑specific compose files.
Observability	Built‑in usage dashboards, token‑level accounting, Stripe billing integration, activity logs.
CLI & Django‑style management	Generated my_ai_app command for server, DB migrations, user creation, etc.
Upgrade path	make upgrade merges upstream template changes into an existing project without losing custom code.
Extensible	Feature flags (e.g., enable_marketing_site) let you add a marketing landing page, SaaS billing, team workspaces, etc.

---

📂 Generated project structure (high‑level)

my_ai_app/
├─ backend/          # FastAPI app, Pydantic models, Celery workers
│   ├─ app/
│   ├─ alembic/      # DB migrations
│   ├─ Dockerfile
│   └─ Makefile
├─ frontend/         # Next.js (React) UI, Tailwind, WebSocket client
│   ├─ pages/        # chat, dashboard, admin, billing, etc.
│   ├─ components/
│   └─ Dockerfile
├─ docker-compose.dev.yml   # dev stack (db, redis, milvus, …)
├─ docker-compose.prod.yml  # production stack
└─ makefile          # shortcuts for dev, prod, upgrade, etc.

The repo also ships a PyPI package (fastapi-fullstack) that contains the scaffolding logic.

---

🤖 AI Agent details

• The agent code lives in the backend and can be selected at generation time (e.g., pydantic_deep).
• Agents can stream responses, call tools, pause for user clarification (ask_user), run Python snippets, and spawn sub‑agents.
• UI renders each tool call as a card, shows a live plan‑and‑task checklist, and visualises charts or code output inline.

---

📚 Documentation & ecosystem

• Docs site – https://vstorm-co.github.io/full-stack-ai-agent-template/ (installation, upgrade guide, API reference).
• Part of the Vstorm OSS ecosystem – linked projects like AgenticOS, pydantic-deepagents, pydantic-ai-shields provide the underlying runtime, guardrails, and sandboxing.
• CI badge shows 100 % test coverage; OpenSSF Best Practices badge indicates a mature open‑source process.

---

🏁 When to use it

• You want a ready‑made SaaS skeleton for an AI‑assistant product (customer support bot, data analyst, code‑assistant, etc.).
• You need enterprise features (auth, multi‑tenant workspaces, Stripe billing) without building them from scratch.
• You prefer a single command to keep the stack up‑to‑date and to upgrade later.

---

📦 How it’s packaged

• The generator is distributed on PyPI as fastapi-fullstack (installable via pip, uv, or pipx).
• The generated app is containerised with Docker and includes optional Kubernetes manifests.

---

📌 TL;DR

vstorm-co/full-stack-ai-agent-template is a genuine, production‑grade project that generates a complete FastAPI + Next.js application pre‑wired for AI agents, RAG, auth, billing, and deployment. It’s aimed at teams building AI‑agent SaaS products and provides a clear upgrade path for the generated code.