Physical oceanography for Open Science Pillars: the ECCO v4r4 state estimate, SWOT KaRIn SSH, meridional transport, budget closure, water masses, sea level, and the PO.DAAC knowledge arc.
This plugin builds on core's foundation skills and its start / discover-data / report workflows, and on the PO.DAAC provider knowledge in nasa-daac-knowledge. It declares both as dependencies, so one install brings them with it.
claude plugin marketplace add open-science-pillars/marketplace claude plugin install ocean-science@open-science-pillars
An install stays at the release it was installed from:
claude plugin update ocean-science@open-science-pillars moves this
plugin to the current one and only this plugin; a dependency moves by
its own update command (claude plugin update core@open-science-pillars,
claude plugin update nasa-daac-knowledge@open-science-pillars), and a
release that raises a floor says so in its notes. claude plugin list
shows what you have. Cowork and Claude Science: add the marketplace and install
from it.
Do the 20-minute ECCO Heat Transport tutorial: it plans the data, loads ECCO through a size gate, computes the Atlantic meridional heat transport at 26.5N, and compares it against the RAPID observations, with every number carrying its uncertainty. New terms are in the glossary.
Every script here (the verification notebooks, the fixture fetch, the
figure renderer) declares its own dependencies in a PEP 723 header and
runs as uv run <script>; uv builds the environment on first run.
The one requirement is uv.
Never python script.py or uv run python script.py: both skip the
header and fail at the first import. uv run <nasa-daac-knowledge>/tools/doctor.py --warm . builds every
environment ahead of a first or offline run.
- Skills for ECCO and SWOT, ocean methods (grids, budgets, transports, water masses, sea level), and workflows: data loaders that show you the download size and ask before fetching, an ocean-budget workflow that refuses to compute on regridded data (it never closes), and transport and comparison workflows that read validated recipes; and a receipt-figures skill that draws maps and series only from attested receipts, with the run id, code hash and verdict in the caption.
- Agents: a scout that recommends datasets and cites the knowledge it relied on (never downloading on its own), and a budget auditor that checks a computed budget closed and diagnoses it if not.
- A knowledge bundle: the plugin's own conventions and connectors.
The PO.DAAC provider bundle (dataset concepts with uncertainty
structure, documented gotchas with evidence, validated recipes with
expected numbers, and attested computations with their sanctioned code
and receipts) is installed alongside as the nasa-daac-knowledge
dependency, at a release the plugin names a floor for. The skills
cite its concepts by bundle path (
knowledge/podaac/...), which core's consult-knowledge convention resolves through the installer's record of installed plugins; nothing is copied into this repository. - Verification: automated notebooks that re-check each workflow on small cached data. The heat-budget closure is additionally attested: a receipt from the sanctioned computation is checked against the code hash and the steward-signed tolerances by ecco-budget-badge (pin tag v2), and the badge above is machine-written from that verdict; it says "does not close" the moment either check fails. Salt, volume, and MHT attach as their computations reach stable. The agent-judgment evals (does an agent with this plugin installed avoid the documented ECCO and SWOT traps?) have one home, ecco-agent-evals: its cases name the skills here they exercise, and each release of this plugin is run against the tagged case set; nothing is copied into this repository.
- A project config template (
ocean-science.local.md.template): copy it into your project and fill in your data paths, region, and download limit.
Downloading NASA data needs an Earthdata Login; nothing else here does,
so you can install the plugin, read the knowledge bundle, and search
collections before you have an account. earthaccess finds the
credential on its own: set EARTHDATA_TOKEN (the easy path in CI and
cloud notebooks), or put a username and password in ~/.netrc
(machine urs.earthdata.nasa.gov, chmod 600), or let it prompt you.
Never commit either to a repo. The
glossary
has the netrc format. License: Apache-2.0. Cite via CITATION.cff.