# Moose / Apis / Openapi Sdk Documentation – Python

## Included Files
1. moose/apis/openapi-sdk/openapi-sdk.mdx

## OpenAPI SDK Generation

Source: moose/apis/openapi-sdk/openapi-sdk.mdx

Generate type-safe client SDKs from your Moose APIs

# OpenAPI SDK Generation

Moose automatically generates OpenAPI specifications for all your APIs, enabling you to create type-safe client SDKs in any language. This allows you to integrate your Moose APIs into any application with full type safety and IntelliSense support.

## Overview

While `moose dev` is running, Moose emits an OpenAPI spec at `.moose/openapi.yaml` covering:

- **Ingestion endpoints** with request/response schemas
- **Analytics APIs** with query parameters and response types

Every time you make a change to your Moose APIs, the OpenAPI spec is updated automatically.

## Generating Typed SDKs from OpenAPI

You can use your preferred generator to create a client from that spec. Below are minimal, tool-agnostic examples you can copy into your project scripts.

### Setup

The following example uses `openapi-python-client` to generate the SDK. Follow the setup instructions here: [openapi-python-client on PyPI](https://pypi.org/project/openapi-python-client/).

Add a generation script in your repository:
```bash filename="scripts/generate_python_sdk.sh" copy
#!/usr/bin/env bash
set -euo pipefail
openapi-python-client generate --path .moose/openapi.yaml --output ./generated/python --overwrite
```

Then configure Moose to run it after each dev reload:
```toml filename="moose.config.toml" copy
[http_server_config]
on_reload_complete_script = "bash scripts/generate_python_sdk.sh"
```

This will regenerate the Python client from the live spec on every reload.

### Hooks for automatic SDK generation
The `on_reload_complete_script` hook is available in your `moose.config.toml` file. It runs after each dev server reload when code/infra changes have been fully applied. This allows you to keep your SDKs continuously up to date as you make changes to your Moose APIs.

Notes:
- The script runs in your project root using your `$SHELL` (falls back to `/bin/sh`).
- Paths like `.moose/openapi.yaml` and `./generated/...` are relative to the project root.
- You can combine multiple generators with `&&` (as shown) or split into a shell script if preferred.

These hooks only affect local development (`moose dev`). The reload hook runs after Moose finishes applying your changes, ensuring `.moose/openapi.yaml` is fresh before regeneration.

## Integration

Import from the output path your generator writes to (see your chosen example repo). The Moose side is unchanged: the spec lives at `.moose/openapi.yaml` during `moose dev`.

## Generators

Use any OpenAPI-compatible generator:

### TypeScript projects
- [OpenAPI Generator (typescript-fetch)](https://github.com/OpenAPITools/openapi-generator) — mature, broad options; generates Fetch-based client
- [Kubb](https://github.com/kubb-project/kubb) — generates types + fetch client with simple config
- [Orval](https://orval.dev/) — flexible output (client + schemas), good DX
- [openapi-typescript](https://github.com/openapi-ts/openapi-typescript) — generates types only (pair with your own client)
- [swagger-typescript-api](https://github.com/acacode/swagger-typescript-api) — codegen for TS clients from OpenAPI
- [openapi-typescript-codegen](https://github.com/ferdikoomen/openapi-typescript-codegen) — TS client + models
- [oazapfts](https://github.com/oazapfts/oazapfts) — minimal TS client based on fetch
- [openapi-zod-client](https://github.com/astahmer/openapi-zod-client) — Zod schema-first client generation

### Python projects
- [openapi-python-client](https://pypi.org/project/openapi-python-client/) — modern typed client for OpenAPI 3.0/3.1
- [OpenAPI Generator (python)](https://github.com/OpenAPITools/openapi-generator) — multiple Python generators (python, python-nextgen)