By the fastCRW team · Catalogue figures checked 2026-11-05 against GET /v1/data · Verify independently before relying on any figure.
The NASA API gives you a small set of open endpoints: near-Earth asteroids, Earth images from the EPIC camera, exoplanet data, and a list of technology projects. Pergamon, the public data catalogue behind fastCRW, puts these behind one API key and the same error and licence shape as every other source. This guide shows the calls that work and the limits you will hit.
What is in the nasa source
The nasa provider exposes 10 capabilities. Ask the find endpoint for the full list with your own key, since the list can change as sources are updated:
curl "https://api.fastcrw.com/v1/data/find?provider=nasa&limit=10" \
-H "Authorization: Bearer $CRW_API_KEY"
The capabilities are asteroid, asteroid-browse, asteroids, asteroid-stats, epic-date, epic-dates, epic-latest, exoplanets, techport-project and techport-projects. The find response lists each one with its required and optional options and its creditsCost.
Near-Earth asteroids
The asteroids capability takes a start_date and an end_date. NASA only serves a feed window of at most 7 days, so keep the two dates 7 days apart or less. On 2026-11-05, a 3-day window (2026-11-01 to 2026-11-03) returned a 200 with an element_count of 16. A 10-day window failed with a 400 from NASA, which fastCRW passed through as upstream_error with the message "NASA returned 400 for this request". That is NASA's cap, not a fastCRW bug, so split the range into 7-day chunks in your loop.
curl "https://api.fastcrw.com/v1/data/nasa/asteroids?start_date=2026-11-01&end_date=2026-11-03" \
-H "Authorization: Bearer $CRW_API_KEY"
For a longer period, let the code split the range. This script yields windows of up to 7 calendar days and makes one call per window, so each request stays inside NASA's cap:
# pip install crw
from datetime import date, timedelta
from crw import CrwClient
def windows(start: date, end: date, span: int = 7):
"""Yield (start, end) date strings, each window at most span calendar days."""
cursor = start
while cursor <= end:
stop = min(cursor + timedelta(days=span - 1), end)
yield cursor.isoformat(), stop.isoformat()
cursor = stop + timedelta(days=1)
client = CrwClient() # reads CRW_API_KEY
for start, end in windows(date(2026, 11, 1), date(2026, 11, 10)):
feed = client.get_data("nasa", "asteroids", {"start_date": start, "end_date": end})
print(start, end, feed["data"]["element_count"])
Run on 2026-11-05, it printed two lines, the same 10-day range that failed above split in two:
2026-11-01 2026-11-07 45
2026-11-08 2026-11-10 17
Each window is one call, so the 10-day range costs two calls. A failed call is not billed, but a failed window still leaves a gap in your data, so log it and retry it later.
Earth images from EPIC
The EPIC camera takes images of Earth. The epic-latest capability needs no options, so it is the simplest call in this guide. On 2026-11-05 it returned a 200 with a list of 13 records, each with keys such as identifier, caption, image, date and centroid_coordinates. The other two EPIC capabilities both require collection, and epic-date also requires date. Read the allowed values from the NASA EPIC documentation.
curl "https://api.fastcrw.com/v1/data/nasa/epic-latest" \
-H "Authorization: Bearer $CRW_API_KEY"
Exoplanets with an ADQL query
The exoplanets capability passes one option, query, to the NASA Exoplanet Archive. That option is an ADQL statement, the SQL-like language the archive's TAP service uses. The Python SDK is the easiest way to send one, because you do not have to URL-encode it:
# pip install crw
from crw import CrwClient
client = CrwClient() # reads CRW_API_KEY
result = client.get_data("nasa", "exoplanets", {
"query": "select top 5 pl_name,disc_year from ps where default_flag=1",
})
print(result["provenance"]["attributionText"])
rows = result["data"] # the archive rows for the query
On 2026-11-05 this query returned 5 rows, each an object with pl_name and disc_year, for example WASP-43 b (2011) and KELT-4 A b (2016). Keep top small while you explore, because a large result is a large response to parse.
TechPort projects
The techport-projects capability needs updatedSince. Without it, fastCRW answers 400 with invalid_options before it calls NASA. Pass a date to pull only the projects NASA changed since then:
curl "https://api.fastcrw.com/v1/data/nasa/techport-projects?updatedSince=2026-10-01" \
-H "Authorization: Bearer $CRW_API_KEY"
For a single project, use the techport-project capability and read its options from the find output.
NASA POWER for climate data
NASA POWER is a separate provider, nasapower, with six capabilities: climatology, daily, hourly, monthly, parameters and regional. The point capabilities need parameters, community, latitude and longitude, and daily, hourly and monthly also need start and end. This call asked for monthly mean air temperature at 2 m (T2M) for the agroclimatology community (AG) near Berlin, for 2023 and 2024, and returned a 200 on 2026-11-05:
curl "https://api.fastcrw.com/v1/data/nasapower/monthly?parameters=T2M&community=AG&latitude=52.5&longitude=13.4&start=2023&end=2024" \
-H "Authorization: Bearer $CRW_API_KEY"
The values sit under data.properties.parameter.T2M, keyed by year and month, for example "202301": 3.18. Each year also has a 13th key (202313), which is NASA POWER's annual value. For other parameter names, call the parameters capability or read the NASA POWER documentation.
Earth Search imagery metadata
The earthsearch provider serves STAC metadata for satellite imagery: collections, and the scenes inside them (collections, collection, collection-items, item, search, aggregate and aggregations). An agent can use it to find out what imagery exists for a place and date. The collections list returned a 200 on 2026-11-05:
curl "https://api.fastcrw.com/v1/data/earthsearch/collections" \
-H "Authorization: Bearer $CRW_API_KEY"
STAC is an open specification for describing geospatial catalogues. See stacspec.org for the item and collection fields. Use limit to keep the list short.
Licence and attribution
Every successful response carries provenance.licence and provenance.attributionText. For the NASA row, the licence line reads: "NASA content is generally not protected by copyright in the US ... NASA logos and insignia are excluded." The full row also says some imagery is third-party and carries its own terms. Keep NASA logos and insignia out of your product, and read the licence text on each response before you ship anything.
The REST response uses snake_case keys, so the attribution is at provenance.attribution_text. The Python SDK returns camelCase, so you read provenance["attributionText"].
Honest limits
- The asteroid feed is capped at 7 days per request, set by NASA.
- NASA can be slow or unavailable. A timeout comes back as
upstream_timeoutwith a 504, and a failed call is not billed. - The exoplanet query language is ADQL, not plain SQL. A query that works in the archive's web form may need small changes.
- Earth Search returns metadata. The image files live at the addresses the metadata points to, and the terms there apply.
- This guide does not cover space weather, launch schedules or mission telemetry. Check the find output for what the catalogue has today.
Sources
- NASA API portal
- NASA Exoplanet Archive: using TAP and ADQL
- NASA POWER documentation
- NASA EPIC
- STAC specification
Related: NASA API on Pergamon · NASA POWER on Pergamon · Earth Search on Pergamon · weather.gov API in Python · Free public APIs for AI agents · Public data API for AI agents · Pergamon catalogue
