fluidkit 1.0.5

Python backend, SvelteKit frontend, zero boilerplate in between.

FluidKit

Web development for the Pythonist

FluidKit bridges Python and SvelteKit into a unified fullstack framework. Write backend functions in Python — FluidKit registers them as FastAPI endpoints and wraps them in SvelteKit-native remote functions with full type safety, cookie forwarding, file uploads, redirects, and single-flight cache invalidation.

pip install fluidkit

How it works

Decorate Python functions. FluidKit registers them as FastAPI endpoints internally and generates colocated .remote.ts files that SvelteKit imports as remote functions directly.

# src/lib/demo.py
from fluidkit import query, command, form

db = {
    "posts": [
        {"id": 1, "title": "Hello World", "content": "This is the first post.", "likes": 10},
        {"id": 2, "title": "Fluidkit", "content": "Fluidkit is awesome!", "likes": 50},
        {"id": 3, "title": "Python and Svelte", "content": "Using Python with Svelte is great!", "likes": 25},
    ]
}

@query
async def get_posts():
    return db["posts"]

@command
async def like_post(post_id: int):
    for post in db["posts"]:
        if post["id"] == post_id:
            post["likes"] += 1

            # invalidates client cache in the same request with single flight mutations
            await get_posts().refresh()
            return True
    return None

@form
async def add_post(title: str, content: str):
    new_post = {
        "id": len(db["posts"]) + 1,
        "title": title,
        "content": content,
        "likes": 0,
    }
    db["posts"].append(new_post)

    await get_posts().refresh() # invalidates client cache in the same request with single flight mutations

<!-- src/routes/+page.svelte -->
<script>
    import { get_posts, like_post, add_post } from '$lib/demo.remote';
</script>

<form {...add_post}>
  <input {...add_post.fields.title.as('text')} placeholder="Title" />
  <input {...add_post.fields.content.as('text')} placeholder="Content" />
  <button>Add Post</button>
</form>

{#each await get_posts() as post}
  <div>
    <h2>{post.title}</h2>
    <p>{post.content}</p>
    <button onclick={async () => await like_post(post.id)}>
      👍 {post.likes}
    </button>
  </div>
{/each}

No manual fetch calls. No duplicated types. No glue code.

🤫 how does this work?

FluidKit reflects on your decorated functions at import time — inspecting parameters, return types, and Pydantic models — and generates colocated `.remote.ts` files wrapping each function in a SvelteKit-native `query`, `command`, `form`, or `prerender` remote function call. In dev mode this re-runs on every save via HMR. The generated files are real TypeScript you can inspect, import, and version control.

Decorators

Decorator	Use case	SvelteKit docs
@query	Read data — cached, refreshable	query
@command	Write data — single-flight cache invalidation	command
@form	Form actions — file uploads, progressive enhancement, redirects	form
@prerender	Build-time data fetching with optional runtime fallback	prerender

CLI

fluidkit init                # scaffold SvelteKit project with FluidKit wired in
fluidkit dev                 # run FastAPI + Vite together with HMR
fluidkit build               # codegen + npm run build
fluidkit preview             # preview production build locally

No system Node.js required — FluidKit uses nodejs-wheel for all Node operations. npm, npx, and node are available through the CLI:

fluidkit install tailwindcss          # shorthand for npm install
fluidkit install -D prettier          # install as dev dependency
fluidkit npm run build                # any npm command
fluidkit npx sv add tailwindcss       # any npx command
fluidkit node scripts/seed.js         # run node directly

Project config

// fluidkit.config.json
{
  "entry": "src/app.py",
  "host": "0.0.0.0",
  "backend_port": 8000,
  "frontend_port": 5173,
  "schema_output": "src/lib/fluidkit",
  "watch_pattern": "./*.py"
}

NOTE: Flags override config. Config overrides defaults.

Built with

• SvelteKit — frontend framework with remote functions
• FastAPI — API layer and request handling
• Pydantic — type extraction and validation
• Jurigged — hot module reloading in dev mode
• nodejs-wheel — bundled Node.js, no system install needed