Thirdtrail Python MCP Server

thirdtrail

Python client for the thirdtrail APIs — sun, moon, twilight, solar irradiance and tides for any point on Earth, with elevation as a first-class input.

pip install thirdtrail

from thirdtrail import Client

tt = Client(api_key="tt_live_…")

for day in tt.ephemerides(lat=53.3498, lon=-6.2603, elevation=20):
    print(day.date, day.sun.rise, day.sun.set)

Why elevation

Most sunrise APIs answer for an observer standing at sea level. The higher you are, the further you see, and the earlier the sun clears the horizon.

At the Cliffs of Moher (214 m) the difference is about three minutes — on sunrise, sunset, and every golden- and blue-hour boundary with them:

Golden hour lasts 20–40 minutes, so three minutes is a real slice of it. Pass elevation whenever you know it.

What you get

day.sun.rise                        # datetime, with its timezone offset intact
day.sun.golden_hour.morning.start
day.moon.phase.name                 # "Waxing Gibbous"
day.moon.phase.illumination         # 0.7376

Timestamps arrive as aware datetime objects, not strings. Anything the client does not model is still there on day.raw and day.response, so a new server-side field never has to wait for a client release.

Endpoints

tt.ephemerides(lat=…, lon=…, elevation=…, days=7)     # sun + moon + twilight
tt.prayer_times(lat=…, lon=…, school="mwl")           # five daily prayers
tt.solar(lat=…, lon=…, timestamp=…)                   # position + irradiance
tt.solar_path(lat=…, lon=…, interval="5m")            # the sun's arc, sampled
tt.solar_dli(lat=…, lon=…, from_date=…, to_date=…)    # Daily Light Integral
tt.tide_heights(lat=…, lon=…, step="10m")             # water height series
tt.tide_extremes(lat=…, lon=…)                        # high and low waters
tt.tide_window(lat=…, lon=…, elevation=…)             # tides + sun + moon
tt.tide_constituents(lat=…, lon=…)                    # harmonic constants

Tides take no elevation — except one

tide_heights and tide_extremes have no elevation parameter, and that is deliberate: how high you stand does not change the ocean surface. Passing it raises TypeError rather than being silently ignored.

tide_window is the exception, because half of what it returns is sun and moon data:

for day in tt.tide_window(lat=52.9715, lon=-9.4309, elevation=214, days=3):
    print(day.date, day.sun.golden_hour.morning.start)
    for t in day.low_waters:
        print("  low water", t.time, t.height_m)

Tide heights are metres relative to mean sea level, not chart datum, and exclude weather-driven surge. Not for navigation.

Errors

Catch a type, not a sentence:

from thirdtrail import PlanLimitExceeded

try:
    tt.tide_extremes(lat=51.5, lon=-2.7, days=90)
except PlanLimitExceeded as e:
    tt.tide_extremes(lat=51.5, lon=-2.7, days=e.limit)   # e.limit == 31

AuthenticationError (401), PermissionDenied (403, a tier you are not on), PlanLimitExceeded and InvalidRequest (400), RateLimited (429), DataUnavailable (503), ServerError (5xx). All subclass ThirdtrailError.

Transient failures and 429s are retried with backoff. A plan ceiling is not — it will fail identically every time.

Keys

Mint one at thirdtrail.life/apis/keys/. A key belongs to one subscription, so it works with one API; a Tides key calling a solar endpoint gets PermissionDenied.

There is a free tier on every API.

Not covered yet

Bulk (POST) endpoints, iCalendar feeds and webhook management. Use the REST API directly for those — the full contract is published as OpenAPI 3.1.

Using an AI agent? There is an MCP server that exposes the same data as tools.

Licence

MIT