siftfy-python

Official Python client for the Siftfy spam-classification API. POST text, get a calibrated spam probability back. One round-trip, no queues, no models to host.

Requirements

Siftfy supports Python 3.9 and newer.

Install

pip install siftfy

Quick start

from siftfy import Siftfy

client = Siftfy(api_key="sk_live_...")

result = client.predict("Win a free iPad — click here!")
print(result.spam_probability)  # 0.97
print(result.likelihood)        # "high"

Get an API key at siftfy.io — the free tier covers 10,000 requests/month at no cost.

Runnable examples

Use these when wiring Siftfy into a real form, signup flow, or moderation queue:

• FastAPI contact-form spam filter
• Next.js route handler
• Django view
• Laravel controller
• Webflow Worker
• Ghost webhook pattern
• Browser spam probability tester

Async

import asyncio
from siftfy import AsyncSiftfy

async def main() -> None:
    async with AsyncSiftfy(api_key="sk_live_...") as client:
        result = await client.predict("hello, world")
        print(result.spam_probability)

asyncio.run(main())

Calibrated probabilities

Every spam_probability is a calibrated value between 0 and 1 — at 0.7, roughly 70% of inputs with that score are actually spam. Pick a threshold appropriate to your use case (a help-desk form tolerates more false positives than a marketplace listing); the same model serves both.

The likelihood field is a coarse bucket ("low", "medium", "high") derived from the probability. Handy for quick branches, but for production decisions thread on the raw probability and your own threshold.

Errors

from siftfy import (
    Siftfy,
    AuthenticationError,  # 401 — bad / revoked key
    RateLimitError,       # 429 — over your tier limit; .retry_after available
    APIError,             # any other 4xx/5xx
    SiftfyError,          # network / request transport errors
)

try:
    result = client.predict(text)
except RateLimitError as e:
    sleep_for = e.retry_after or 1.0
    ...
except AuthenticationError:
    ...
except APIError as e:
    log(f"siftfy error {e.status_code}: {e} (request_id={e.request_id})")

The client retries idempotent failures (HTTP 408 / 429 / 5xx, network errors) with exponential backoff and jitter, honouring Retry-After when present. Tune with max_retries=N (default 2; set 0 to disable).

Configuration

client = Siftfy(
    api_key="sk_live_...",
    base_url="https://api.siftfy.io",  # override for self-hosted / staging
    timeout=10.0,                       # seconds, applied per attempt
    max_retries=2,                      # 0 disables retries
)

You can also pass your own httpx.Client (or httpx.AsyncClient for the async client) via http_client=... if you want connection pooling, custom transports, or to share a client across services.

Development

python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
pip install -e ".[dev]"

ruff check src tests
mypy src
pytest -q
python -m build

Support and security

Use GitHub Issues for SDK bugs and feature requests. Do not open public issues for suspected vulnerabilities; report them privately using the process in SECURITY.md.

Resources

• API reference: https://siftfy.io/docs
• Predict endpoint: https://siftfy.io/docs/predict
• Runnable examples: https://siftfy.io/examples
• Free anti-spam tools: https://siftfy.io/tools
• Contact-form guide: https://siftfy.io/use-cases/contact-forms
• Comparison guide: https://siftfy.io/best-spam-detection-api
• Pricing: https://siftfy.io/pricing
• Status: ping https://api.siftfy.io/health
• Issues: https://github.com/siftfy/siftfy-python/issues

License

MIT — see LICENSE.