A small CircuitPython 9.x library for sending custom metrics to the Datadog HTTP API at POST /api/v2/series.

The library is designed for network-capable CircuitPython boards, including boards with built-in WiFi, Ethernet, or a supported network coprocessor. It has no runtime dependencies beyond time and the adafruit_requests.Session object supplied by your own network setup code.

Copy datadog.py to the lib/ folder on your CircuitPython device:

CIRCUITPY/
|-- code.py
`-- lib/
    `-- datadog.py

This repository follows the usual single-file CircuitPython library layout:

circuitpython-datadog/
|-- datadog.py
|-- examples/
|   `-- basic_metrics.py
|-- tests/
|   `-- test_datadog.py
|-- README.md
`-- LICENSE

You are responsible for installing adafruit_requests and its required support files on the device, and for connecting the network before creating the client.

import ssl
import socketpool
import wifi
import adafruit_requests
from datadog import DatadogClient

# This setup is for boards with built-in WiFi. If your board uses Ethernet
# or an external coprocessor, create the session for that network interface.
wifi.radio.connect("ssid", "password")
pool = socketpool.SocketPool(wifi.radio)
session = adafruit_requests.Session(pool, ssl.create_default_context())

client = DatadogClient(
    session,
    "DATADOG_API_KEY",
    site="datadoghq.com",
    default_tags=["runtime:circuitpython", "board:my-board"],
)

client.gauge("circuitpython.temperature", 24.5, tags=["room:lab"])
client.count("circuitpython.loop.count", 1)

if not client.flush():
    print("Datadog flush failed; continuing")

See examples/basic_metrics.py for a complete code.py style example.

DatadogClient(
    session,
    api_key,
    site="datadoghq.com",
    default_tags=None,
    max_retries=3,
    retry_delay=2,
)

Parameters:

• session: an already configured adafruit_requests.Session.
• api_key: your Datadog API key string.
• site: Datadog site string used to build the intake host. The default is datadoghq.com, which sends to https://api.datadoghq.com/api/v2/series.
• default_tags: optional list of Datadog tag strings applied to every metric.
• max_retries: number of retry attempts after a failed request. The default is 3, so flush() may try up to four total HTTP requests.
• retry_delay: seconds to wait between retry attempts using time.sleep().

The client buffers metrics in RAM until flush() is called:

client.gauge("sensor.temperature", 23.8)
client.count("button.press", 1, tags=["button:a"])
client.rate("sensor.samples_per_second", 2.0)
client.flush()

Pass timestamp= to use an explicit Unix timestamp instead of time.time():

client.gauge("sensor.temperature", 23.8, timestamp=1710000123)

flush() returns True when Datadog accepts the payload with a 2xx HTTP status. It returns False on network errors, socket errors, HTTP error status codes, or unexpected request failures. It never intentionally raises an exception, and it clears the buffer after permanent failure to avoid memory buildup in sensor loops.

Each request response is closed with response.close().

The core client can be tested on desktop Python because network access is injected through the session object:

python3 -B -m unittest discover -s tests

The tests use fake sessions and responses. They validate payload shape, retry behavior, response cleanup, URL construction, and buffer clearing without making real network requests.

Pass the site hostname for your Datadog account:

• datadoghq.com
• datadoghq.eu
• us3.datadoghq.com
• us5.datadoghq.com
• ap1.datadoghq.com
• ap2.datadoghq.com
• ddog-gov.com

The client prepends api. when needed, so us3.datadoghq.com becomes https://api.us3.datadoghq.com/api/v2/series.

• Network setup, TLS setup, NTP/RTC time setup, and the requests session are caller responsibilities.
• Datadog expects metric timestamps to be close to current Unix time. If your board does not have correct time after boot, set the RTC before sending metrics.
• This library does not use threading, asyncio, subprocesses, or environment variables.