How AI agents follow ESA and European space missions
An agent follows European space activity by calling a wire that reads the European Space Agency's news feed at runtime. It returns mission milestones, Earth observation results and programme decisions. ESA is an intergovernmental agency, separate from the European Union, and conflating the two produces a wrong attribution.
/mcp. Nothing here needs an API key or a signup.
Two mistakes account for most bad answers about European space, and both are attribution mistakes rather than data problems.
ESA is not the EU, and it is not a European NASA
ESA is an intergovernmental organisation with 23 Member States, a 2025 budget of €7.68 billion and roughly 3,000 staff. Its membership does not match EU membership: the United Kingdom, Norway and Switzerland are ESA Member States and are not EU members. So "the EU launched" is wrong for most ESA missions, and an agent that treats the two as interchangeable will misattribute both directions.
The relationship runs the other way for some programmes. Copernicus and Galileo are EU programmes, funded by the EU, with ESA acting as the technical and procurement agent. A Copernicus satellite is an EU asset built and operated through ESA. Getting that right in one sentence is usually enough:
| Programme | Owner | ESA's role |
|---|---|---|
| Copernicus, Galileo | European Union | Development and procurement agent |
| ExoMars, Euclid, JUICE | ESA | Owner and operator |
| Webb, Hubble | NASA-led partnerships | Contributing partner |
The second mistake is treating ESA as the European equivalent of NASA in operations. NASA operates its own launch programmes directly. ESA's launchers are flown commercially by Arianespace, and ESA is a customer and a development funder rather than the launch operator. An answer that says "ESA launched the rocket" is loose in a way that matters to anyone tracking the actual supply chain. Policy and funding decisions taken at EU level are a different artefact again, covered in how agents follow European Commission decisions.
The feed quirk that fakes freshness
A live read of the ESA feed returned 9 items. Eight were from the previous week: an ExoMars milestone, a Webb result, a call for software to fly on an asteroid mission, a forest biomass finding, and a World Space Week item. The ninth was a newsletter entry dated 21 January 2026, sitting in the same feed at the same position weight as items from yesterday.
That is the failure mode. An agent that reads the feed top-down, or that trusts feed order as recency order, will present a six-month-old item beside a one-day-old one. Every item on this wire carries its published timestamp and an age, and since filters on that timestamp rather than on position. Retention is 14 days with a 14 day default lookback, so the January item is outside the window entirely on a default call. The general version of this problem, and why undated feeds are refused at ingest, is in how agents follow new arXiv papers.
Volume is the other thing to plan for. ESA Top News publishes a handful of items a week, so a 24-hour window returns empty on most days and reads to an agent like a broken route. The 14 day default exists for that reason.
The call an agent makes
# discovery only, no items: browse every wire, its filters and its price claude mcp add --transport http botwire https://thebotwire.com/mcp # paid, $0.005: ESA mission and programme news across the window curl -i "https://thebotwire.com/esa/latest?q=ExoMars&since=14d&limit=25" # NASA is a separate wire, same query grammar curl -i "https://thebotwire.com/space/latest?q=asteroid&since=14d"
Parameters are q, since and limit (1 to 50). One publisher sits behind the ESA wire, so src accepts a single value and does no useful work. The call costs $0.005 in USDC on Base, with no API key, no signup and no subscription. Mission names, instrument names and spacecraft names outperform a phrased question, because search here is ranked keyword matching. Every route, filter and price is documented at /docs.
FAQ
Does this wire include launch schedules?
No. It carries announcements about missions, including news that a launch happened or has been scheduled, but it is not a launch manifest. An agent asked for the next European launch window should say it is reporting an announcement rather than a schedule, and route to the launch provider's own manifest when the date has to be exact.
Is ESA science data available through this wire?
No. The wire carries news releases about results. The underlying datasets from missions such as Gaia, Euclid and the Copernicus Sentinels are published through ESA's own archives and the Copernicus data services, which are separate systems with their own access terms. The news item is the pointer, not the data.
Why is ESA membership different from EU membership?
Because ESA was founded as an independent intergovernmental agency by treaty, not as an EU institution. States join by acceding to the ESA Convention and paying into its programmes, which is why the UK, Norway and Switzerland are members while several EU states participate under different arrangements. The practical effect for an agent is that "European" is ambiguous and the specific organisation should always be named.
How fresh is the ESA wire?
It sits in the sparse ingest tier and is polled at most every 15 minutes. The binding constraint on this route is the publication schedule, not the poll interval, because ESA publishes a few items a week. The observed ingest interval and the last poll timestamp are published live at /health.
Sources
ESA's membership of 23 Member States, its 2025 budget of €7.68 billion and its staff of roughly 3,000 are stated by the agency on its ESA facts page. The ESA news feed used by this wire is the agency's own space news feed, which returned 9 items when read on 7 August 2026. Poll cadence and per-call pricing for this service are published live at thebotwire.com/health.
Related: How do AI agents follow new arXiv papers? · How do AI agents follow European Commission decisions?