floopfloop

Official Python SDK for the FloopFloop API. Build a project, refine it, manage secrets and API keys from any Python 3.10+ codebase.

Install

pip install floopfloop

Quickstart

Grab an API key via the floop CLI (floop keys create my-sdk) or the dashboard → Account → API Keys. Business plan required to mint new keys.

import os
from floopfloop import FloopClient

floop = FloopClient(api_key=os.environ["FLOOP_API_KEY"])

# Create a project and wait for it to go live.
created = floop.projects.create(
    prompt="A landing page for a cat cafe with a sign-up form",
    name="Cat Cafe",
    subdomain="cat-cafe",
    bot_type="site",
)
live = floop.projects.wait_for_live(created["project"]["id"])
print("Live at:", live["url"])

Streaming progress

for event in floop.projects.stream(project_id):
    print(f"{event['status']} ({event['step']}/{event['total_steps']}) — {event['message']}")

Error handling

Every call raises FloopError on non-2xx. Switch on .code:

import time
from floopfloop import FloopClient, FloopError

try:
    floop.projects.create(prompt="...")
except FloopError as err:
    if err.code == "RATE_LIMITED":
        time.sleep((err.retry_after_ms or 5000) / 1000)
    elif err.code == "UNAUTHORIZED":
        print("Check your FLOOP_API_KEY.")
    else:
        print(f"[{err.request_id}] {err.code}: {err}")
    raise

Resources

For longer end-to-end patterns — streaming a build, refining mid-deploy, attachment uploads, key rotation, retry-with-backoff — see the cookbook.

Authentication

Two token shapes are accepted on api_key:

CLI device tokens (flp_cli_…) work here too — handy for local scripts that already logged in through the CLI.

Configuration

floop = FloopClient(
    api_key="flp_...",
    base_url="https://www.floopfloop.com",   # override for staging
    timeout=30.0,                             # seconds
    poll_interval=2.0,                        # for wait_for_live / stream
    user_agent="myapp/1.2.3",                 # appended to User-Agent
)

Use as a context manager to close the underlying httpx.Client:

with FloopClient(api_key="...") as floop:
    floop.projects.list()

Async

AsyncFloopClient mirrors FloopClient on top of httpx.AsyncClient — same arguments, same resource accessors (projects, secrets, api_keys, ...), every method is async def. Use async with to ensure the underlying HTTP client is closed:

import asyncio
from floopfloop import AsyncFloopClient

async def main():
    async with AsyncFloopClient(api_key="flp_...") as floop:
        created = await floop.projects.create(prompt="a cat cafe landing page")
        live = await floop.projects.wait_for_live(created["project"]["id"])
        print(live["url"])

        # Streaming uses `async for`:
        async for ev in floop.projects.stream(created["project"]["id"]):
            print(ev["status"], ev.get("progress"))

asyncio.run(main())

Releasing (maintainers only)

This package publishes to PyPI via trusted publishing (OIDC) — no long-lived API token is stored in GitHub Actions.

Before the first release, register the trusted publisher at https://pypi.org/manage/project/floopfloop/settings/publishing/ (or, for a brand-new project, the pending-publisher page):

• PyPI project name: floopfloop
• Owner: FloopFloopAI
• Repository name: floop-python-sdk
• Workflow name: release.yml
• Environment name: leave blank

After that, ship a release by bumping pyproject.toml + src/floopfloop/_version.py, pushing a py-v<version> tag. The release workflow will build, test, and publish.

License

MIT. See LICENSE.

Related

• @floopfloop/sdk — Node.js twin of this SDK; same backend surface.
• floop-cli — the CLI that ships the same backend surface as a terminal UI.
• Customer docs — API reference.