A Python client for the SolarEdge Monitoring API (V2), with synchronous and asynchronous interfaces and a method for every documented operation.
Upgrading from 2.x? SolarEdge retired Monitoring API V1 on November 3, 2026, and every request from 2.x now fails with HTTP 410. 3.0 talks to V2, which needs a new API key and has different endpoints and response shapes. See Migrating from 2.x.
- Features
- Installation
- Credentials
- Quick Start
- Rate Limits
- Migrating from 2.x
- Development
- API Documentation
- Complete V2 coverage: all 59 operations in SolarEdge's Basic Monitoring
(2.2.0) and Advanced Monitoring (2.0.1) APIs, plus
get()for anything added later - Sync & Async Support:
MonitoringClientandAsyncMonitoringClientshare one definition of every method, so they cannot drift apart - Typed: full annotations,
py.typed,Literaltypes for every enumerated parameter, and aTypedDictfor every response, generated from SolarEdge's specs - OAuth built in: Site Access tokens are refreshed automatically, with a
uv run python -m solaredge loginhelper for homeowners - Safe by default: arguments are validated before anything is sent, and credentials never appear in URLs, reprs or error messages
uv add solaredgeV1 API keys do not work with V2. Which V2 credential you need depends on who you are:
| You are | Access type | Pass to the client |
|---|---|---|
| A homeowner reading your own system | Site Access (OAuth 2.0) | oauth=OAuth2(...), or access_token="..." |
| An installer with a fleet of sites | Fleet Access (API key) | api_key="..." |
SolarEdge only offers Fleet Access keys to installer accounts, so homeowners use OAuth. See SolarEdge's Homeowner and Self-Access Guide.
-
Create a developer account: sign in to the Developer Platform with your mySolarEdge login and fill in the developer profile. The Free tier is enough.
-
Create a Site Access application:
- Service: Monitoring API.
- Redirect URI:
http://localhost:8080/oauth/callback. - Scopes:
SITE_DATAandDEVICE_DATA. - Copy the client secret: it is shown only once.
-
Authorize the app for your site with the bundled helper. It opens the consent page, catches the redirect and saves the token to a file only you can read:
export SOLAREDGE_CLIENT_SECRET=... # or omit it to be prompted uv run python -m solaredge login --client-id YOUR_CLIENT_ID # Saved the token for site 1234567 to solaredge-token.json (mode 600).
-
Use the token. Access tokens last 2 hours;
OAuth2refreshes them automatically. Each refresh also replaces the refresh token, so save every new token inon_token_update, or the next run will hold a dead one. A refresh token unused for 14 days expires, and you have to runloginagain.
import json
from pathlib import Path
from solaredge import MonitoringClient, OAuth2, OAuthToken
TOKEN_FILE = Path("solaredge-token.json")
def save(token: OAuthToken) -> None:
TOKEN_FILE.write_text(json.dumps(token.to_dict()))
auth = OAuth2(
OAuthToken.from_dict(json.loads(TOKEN_FILE.read_text())),
client_id="YOUR_CLIENT_ID",
client_secret="YOUR_CLIENT_SECRET",
on_token_update=save,
)
with MonitoringClient(oauth=auth) as client:
overview = client.get_overview(auth.token.site_id)OAuth2 works with both clients. With AsyncMonitoringClient,
on_token_update may be an async function. Concurrent requests share a single
refresh, so they never spend the same refresh token twice.
If you already manage tokens yourself, for example through Home Assistant's
OAuth helper, pass the current one with MonitoringClient(access_token="...").
Building an app for other homeowners? solaredge.oauth has the pieces:
authorization_url, parse_callback, exchange_code, refresh_token,
revoke_token, and their async_ variants.
Create a Fleet Access application on the Developer Platform and pass its key:
MonitoringClient("YOUR_API_KEY"). It is sent in the X-API-Key header, and
it is the only credential accepted by get_sites and get_alerts, which list
data across a fleet.
All time windows take timezone-aware datetimes, which are sent as UTC. V2
rejects timestamps without an offset, so a naive datetime raises
SolarEdgeValidationError before any request is sent. Where start/end are
optional, leaving them out means "midnight today until now" in the site's
time zone.
The examples below use auth, the OAuth2 object from
Credentials. With a Fleet Access key, pass
api_key="..." instead.
from datetime import datetime, timedelta, timezone
from solaredge import MonitoringClient
SITE_ID = 1234567
now = datetime.now(timezone.utc)
with MonitoringClient(oauth=auth) as client:
details = client.get_site_details(SITE_ID)
today = client.get_overview(SITE_ID)
daily = client.get_energy(
SITE_ID,
start=now - timedelta(days=30),
end=now,
resolution="DAY",
unit="KWH",
)
batteries = client.get_storage_telemetry(SITE_ID)import asyncio
from datetime import datetime, timedelta, timezone
from solaredge import AsyncMonitoringClient
async def main():
now = datetime.now(timezone.utc)
async with AsyncMonitoringClient(oauth=auth) as client:
details, power = await asyncio.gather(
client.get_site_details(1234567),
client.get_power(1234567, start=now - timedelta(hours=2), end=now),
)
asyncio.run(main())Every method returns the parsed JSON unchanged, typed as a TypedDict (see
Typed Responses). Optional arguments left as
None are not sent, so the API's own defaults apply. The
API reference lists the paths, parameters and
per-resolution window limits.
| Area | Methods |
|---|---|
| Sites | get_sites, get_site_details, get_overview, get_performance, get_power, get_energy, get_environmental_benefits, get_lifetime_energy, get_external_production_energy |
| Alerts | get_alerts (all sites), get_site_alerts |
| Inventory | get_devices, get_device |
| Telemetry | get_inverters_telemetry, get_meters_telemetry, get_storage_telemetry, get_ev_chargers_telemetry, get_sensors_telemetry |
| Status | get_inverters_status, get_storage_status |
| Lifetime counters | get_inverters_lifetime_energy, get_meters_lifetime_energy, get_storage_lifetime_energy |
| Inverter | get_inverter_power, _energy, _consumed_ac_power, _consumed_ac_energy, _voltage, _current, _frequency |
| Meter | get_meter_production_power / _energy, get_meter_consumption_*, get_meter_import_*, get_meter_export_* |
| Battery | get_battery_charge_power, _discharge_power, _charge_energy, _discharge_energy, _state_of_energy, _remaining_energy |
| Advanced Monitoring* | get_power_per_destination, get_energy_per_destination, get_current_power_flow, start_live_power_flow, get_weather_guard, get_backup_reserve |
| Anything else | get(path, params=None) |
* Business Pro and Enterprise plans only. Other plans get HTTP 403, raised as
SolarEdgeAuthError.
Telemetry, status and lifetime methods cover every device of that kind on the
site. Pass serial_numbers=[...] to limit them to specific devices (sensors,
which have no serial number, are always returned together). V2 reports
units in upper case ("WH", "KWH", "W"), so compare them
case-insensitively.
Responses are annotated with TypedDicts generated from SolarEdge's OpenAPI
specs, so your editor completes field names and a type checker catches a
misspelt key. The async client returns awaitables of the same types.
from solaredge.types import Measurements
power: Measurements = client.get_power(SITE_ID)
for sample in power.get("values", []):
print(sample.get("timestamp"), sample.get("value"))They are type hints only: at runtime every response is still a plain dict.
Fields the spec doesn't mark as required are NotRequired, because real
responses can leave them out, so prefer .get() for those. All the types live
in solaredge.types.
Every failure raises a SolarEdgeError subclass, so you never need to import
httpx to handle one. The originating httpx exception is preserved as
__cause__ if you want the transport-level detail.
from solaredge import (
MonitoringClient,
SolarEdgeAuthError,
SolarEdgeRateLimitError,
SolarEdgeError,
)
with MonitoringClient("YOUR_API_KEY") as client:
try:
flow = client.get_current_power_flow(1234567)
except SolarEdgeAuthError:
... # bad credentials, or an operation your plan does not include
except SolarEdgeRateLimitError as exc:
... # exc.retry_after, exc.limits
except SolarEdgeError:
... # anything else this library raisesThe hierarchy:
| Exception | Raised when |
|---|---|
SolarEdgeError |
Base class for everything below |
SolarEdgeValidationError |
Arguments rejected locally, before any request. Also a ValueError |
SolarEdgeAPIError |
The API returned an error status, e.g. 400 for a window that is too long |
SolarEdgeAuthError |
401 / 403: credentials invalid or expired, missing scope, or the operation is not in your plan |
SolarEdgeOAuthError |
The token endpoint rejected a code or refresh token. Carries error and error_description; an expired grant means running login again |
SolarEdgeNotFoundError |
404: no such site, device or operation |
SolarEdgeRateLimitError |
429: per-minute or monthly quota exceeded. Carries retry_after and limits (the x-ratelimit-* headers) |
SolarEdgeServerError |
5xx: the API failed to handle a valid request |
SolarEdgeResponseError |
Success status, but the body could not be parsed |
SolarEdgeAPIError and its subclasses carry status_code, response_body and
url.
Every call costs one credit. Plans set a monthly credit quota and a calls-per-minute limit; the Free tier allows 2,000 credits a month and 10 calls a minute.
This client caps itself at 3 concurrent requests but does not retry or throttle for you. A client that polled several endpoints every 15 minutes under V1 will run through the Free tier's monthly quota. To stay within it:
- Prefer the bulk telemetry methods over per-device ones.
- Use coarser resolutions where you can.
- Poll less often.
| 2.x method | 3.0 replacement |
|---|---|
get_site_list(...) |
get_sites(page=..., sites_in_page=..., name=...) |
get_site_details(site_id) |
get_site_details(site_id) |
get_overview(site_ids) |
get_overview(site_id), one site per call |
get_energy(site_ids, start_date, end_date, time_unit) |
get_energy(site_id, start, end, resolution, unit) |
get_time_frame_energy(...) |
get_energy(..., resolution="TOTAL") |
get_power(site_id, start_time, end_time) |
get_power(site_id, start, end, resolution, unit) |
get_power_details(...), get_energy_details(...) |
get_meters_telemetry(...), or get_power_per_destination / get_energy_per_destination (Advanced) |
get_current_power_flow(site_id) |
get_current_power_flow(site_id) (Advanced) |
get_storage_data(...) |
get_storage_telemetry(...), get_storage_status(...) |
get_environmental_benefits(site_id, system_units) |
get_environmental_benefits(site_id, unit="METRIC" or "IMPERIAL") |
get_inventory(site_id), get_components_list(site_id) |
get_devices(site_id, types=[...]) |
get_inverter_technical_data(site_id, serial, ...) |
get_inverters_telemetry(...) and get_inverters_status(...) with serial_numbers=[serial] |
get_meters(...) |
get_meters_lifetime_energy(...), get_meters_telemetry(...) |
get_site_data(site_ids) |
No V2 equivalent |
Other changes:
- Credentials: V1 keys are rejected. Homeowners use Site Access (OAuth); installers use a V2 Fleet Access API key. See Credentials.
- One site per call: V2 has no comma-separated bulk form for sites.
- Datetimes must be timezone-aware, and are sent as UTC
from/to. - Parameters renamed:
time_unitis nowresolution, andQUARTER_OF_AN_HOURis nowQUARTER_HOUR. Enum values are upper case, e.g.METRIC,PRODUCTION. - Response bodies differ, so code that reads fields needs updating.
- Data no longer available: optimizer-level telemetry (optimizers appear
only in
get_devices), and per-phase inverter values (V2 averages across phases).
# Install development dependencies
uv sync
uv run prek install -t commit-msg# Committing changes
uv run cz c# Regenerate solaredge/types.py from the specs in specs/
uv run --group codegen scripts/gen_types.py
# ...or download SolarEdge's latest published specs first
uv run --group codegen scripts/gen_types.py --fetchReleases are automatic. Every merge to main runs the full test suite, then
commitizen reads the conventional commits since the last tag:
fix:releases a patch,feat:a minor version, andBREAKING CHANGE(orfeat!:) a major version.docs:,ci:,test:,chore:and the like don't release anything.
When there's something to release, CI commits the version bump and CHANGELOG
to main, creates the tag and GitHub release, and publishes to PyPI.
- The PR title decides the release. PRs are squash-merged, so the title
becomes the commit commitizen reads. Give it the right type, and mark breaking
changes with
!(e.g.feat!: ...) or aBREAKING CHANGE:footer in the squash message. A CI check rejects titles that aren't conventional commits. To merge without releasing, use a non-release type likedocs:orci:. - Don't bump or tag by hand. No local
cz bump, no bump PRs, no manual tags or GitHub releases. CI picks the version. - If a release fails partway, e.g. PyPI was down after the tag was created, use Re-run failed jobs on that run. Re-run all jobs sees the tag already exists and skips publishing.