Python quickstart

reconify-python is a server-side client for Python 3.10 or newer. It needs a write or admin key, which Authentication covers.

Reconify has no sandbox or test mode. The event you send below becomes immutable evidence in your real organization, and the public API deletes nothing.

Install

The distribution installs from PyPI:

pip

pip install reconify-python

The distribution is reconify-python, and the import name is reconify.

Configure a client

A client holds the key and the base URL:

import os

from reconify import Reconify

client = Reconify(api_key=os.environ["RECONIFY_API_KEY"])

The client reads RECONIFY_API_KEY and RECONIFY_API_URL when you omit api_key and base_url. The default base URL is https://api.reconifyhq.com/v2, and the client appends /v2 when your value omits it.

The constructor raises ReconifyValidationError when the key is missing or does not start with rk_. This happens before any network call, so a misconfigured deployment fails at startup rather than at the first event.

Both clients hold an HTTP connection pool, so they work as context managers. Outside a context manager, Reconify needs close() and AsyncReconify needs await aclose().

Send one event

Request fields use the wire format, and amount is a decimal string such as "150.00" rather than a float:

from datetime import datetime, timezone

from reconify import Reconify
from reconify.models import MonitoringBatchRequest, MonitoringEvent

batch = MonitoringBatchRequest(
    events=[
        MonitoringEvent(
            id="evt_01J3Y0M8VJQ5W1R3E4J4K7N8P9",
            flow="payment_to_wallet",
            type="payment.succeeded",
            occurred_at=datetime(2026, 1, 1, 12, 0, tzinfo=timezone.utc),
            amount="150.00",
            currency="USD",
            reference="order-123",
            entity_id="wallet_123",
        )
    ]
)

with Reconify() as client:
    result = client.ingestion.ingest_monitoring_events(batch)

    for item in result.results:
        print(item.index, item.status, item.event_id or item.code)

Each result is accepted, duplicate, or rejected, and its index matches the position of the event you sent. accepted means Reconify stored the event. It does not mean the monitored operation is complete.

Field names

Model fields use the wire format, not Python casing conventions applied to something else: send occurred_at, entity_id, and type. amount is a decimal string, such as "150.00". Passing a float fails model validation. Format money from Decimal, never from float arithmetic. One name differs between request and response: you send type, and an event read returns event_type. Request models reject unknown fields before sending. That turns a typo into an immediate ReconifyValidationError instead of a rejected event.

Use the async client

AsyncReconify exposes the same modules and method names.

import asyncio

from reconify import AsyncReconify

async def main() -> None:
    async with AsyncReconify() as client:
        result = await client.ingestion.ingest_monitoring_events(batch)
        print(result.results)

asyncio.run(main())

Read the event back

A read confirms what Reconify stored:

with Reconify() as client:
    event = client.events.get_event("evt_01J3Y0M8VJQ5W1R3E4J4K7N8P9")
    print(event.event_type, event.status)

Build batches, handle every item result, and retry safely in Send events with Python.

Overview

TypeScript

Python

Install

Configure a client

Send one event

Field names

Use the async client

Read the event back

Next

​Install

​Configure a client

​Send one event

​Field names

​Use the async client

​Read the event back

​Next

Install

Configure a client

Send one event

Field names

Use the async client

Read the event back

Next