Test-only full-content Orinoco Lite downstream for the Center for Open Neuroscience.
This is an ordinary single-repository Orinoco Lite consumer. Metadata records, editorial content, declared assets, presentation, source adapters, extensions, and site tests are versioned directly here. Building, previewing, and updating it requires neither Git submodules nor an engineering workspace checkout.
- This repository owns its content, presentation, policy, tests, review, and deployment; if present,
site/README.mdis its site-owned guide for concrete editorial and operating procedures. - The Orinoco Lite template owns the generic repository facade, file-ownership contract, and content-preserving updater.
- The Orinoco Lite engine implements the commands, runtime verification, projection, and static build.
orinoco.lock is the release authority.
It records exact engine, runtime, template, and reusable-workflow coordinates; the frozen pixi.lock realizes that reviewed environment.
PyPI publication is optional and is not required by this repository.
Start with creation and configuration for the required site-profile step, file ownership before customizing the facade, and framework updates before changing release pins.
The generic Orinoco Lite facade is MIT licensed and its original documentation is CC BY 4.0. Those terms do not license site-owned records, editorial prose, media, branding, presentation, or imported third-party material. Document those rights separately and preserve every upstream notice. See engine human-review decision HR-003.
pixi run validate
pixi run projection-update
pixi run projection-verify
pixi run assets-hydrate
pixi run assets-verify
pixi run build
pixi run serve
pixi run test
pixi run test-all
pixi run update-checkAfter editing metadata records, validate regenerates the ignored projection and checks it.
build does the same before rendering, so the source commit shows the metadata change rather than a duplicate generated tree.
assets-hydrate is the explicit networked retrieval step for declared remote assets; assets-verify checks already-local payloads without fetching.
pixi run build writes build/site with root-relative links.
pixi run serve serves that existing artifact on port 8765; it does not rebuild it.
The same files therefore work at both http://127.0.0.1:8765/ and http://localhost:8765/.
Pages is intentionally separate.
The Pages workflow obtains the destination's absolute public base URL from GitHub, validates it, and passes it to pixi run build-pages, which writes build/pages.
Browser acceptance uses a controlled local project-path URL matching the repository slug.
Neither target changes the canonical public identity recorded in orinoco.yaml.
pixi run test-all is the complete acceptance gate: asset preparation, configuration and runtime validation, projection verification, the pinned Hugo Extended version, consumer tests, byte-compared repeat builds, dual-loopback local-link checks, and the checked Chromium/WebKit scenarios.
Its browser preparation makes both engines available before testing; on Linux, the post-Chromium WebKit host-library step and browser downloads are separate, logged, bounded phases.
Hydration is the only asset command authorized to retrieve declared read-only payload URLs.
For a warmed-cache offline proof, run assets-hydrate while online, deny network access at the operating-system boundary, then run assets-verify before offline validation, projection, build, and editor checks.
Do not use assets-prepare-online in that denied-network phase; it represents the normal cold-clone preparation path.
metadata/records/contains every YAML Thing used as projection input..orinoco-lite/contains implementation support behind the checked commands.custom/editorial/,custom/assets/, andsite/contain site-owned presentation inputs..agents/skills/manage-orinoco-content/guides agents through focused editorial and asset changes..agents/skills/operate-orinoco-metadata-adapters/guides adapter runs, explicit human decisions, provenance, and review pull requests.source-adapters/contains optional site-owned importers, enrichers, and scrapers; it is not a deployed runtime dependency.extensions/is the stable downstream customization surface.generated/contains ignored projection output recreated by validation and builds.
A newly created repository is a content-neutral facade, not an empty but buildable website. Add a reviewed site profile as described in creation and configuration before running validation or build commands. A populated profile and source-adapter examples are available in the downstream test website.