---

name: metadata-and-catalogue description: "Operate GeoNode metadata schemas, handlers, localization, catalogue/CSW publication, search filters, facets, and metadata-adjacent preview context without confusing data validation with service availability." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

Metadata and catalogue

Use this sub-skill when a task edits, validates, localizes, indexes, searches, or publishes resource metadata, or when a catalogue/search result does not match expectations. It covers datasets and other ResourceBase resources. Route upload mechanics and ordinary resource CRUD to resource-and-api; route GeoServer publication, service URLs, styles, and permissions to geoserver-and-security; route remote harvesting and worker recovery to harvesting-and-admin.

Operating boundary

• Treat the metadata JSON Schema as the authoritative shape for the metadata editor/API instance. Do not invent a replacement ISO, FGDC, or Dublin Core schema.
• Treat the configured catalogue backend and its URL as separate service dependencies. A valid local document does not prove CSW availability.
• Treat required metadata annotations as editor guidance unless the active deployment adds a separate policy gate: the user guide says incomplete metadata remains usable.
• Never send credentials, mutate a remote catalogue, or run a network CSW helper from the bundled validator.
• External PostgreSQL/PostGIS, GeoServer, Redis/Celery, remote CSW/OGC endpoints, and browser services are prerequisites for corresponding live workflows and are not verified by this skill.

Route the request

1. Identify the resource primary key or UUID and whether the request is about schema, one resource instance, XML import/export, catalogue transport, search, facets, or preview/style context.
2. For schema/instance work, read metadata-reference.md.
3. For CSW, OpenSearch, catalogue links, full-text search, or facets, read catalogue-and-search.md.
4. For a supplied local JSON/XML file, run scripts/validate-metadata.py first; use explicit required fields when a deployment or task supplies them.
5. If the validator passes but a catalogue request fails, switch to the service-gate diagnosis in troubleshooting.md rather than changing the metadata blindly.

Metadata API workflow

• Discover the metadata endpoints from the API base: GET /api/v2/metadata/ returns schema and instance URL templates.
• Fetch the localized schema with GET /api/v2/metadata/schema/?lang=<2-letter>. Record the returned properties, top-level required, each geonode:handler, and any autocomplete URL before constructing a payload.
• Fetch an instance with GET /api/v2/metadata/instance/<pk>/?lang=<lang>. Preserve object/array shapes and IDs returned by the server.
• Use PUT for a complete instance or PATCH for selected fields. A successful update returns HTTP 200; handler/save/index errors return HTTP 422 with message and nested extraErrors. A missing resource is HTTP 404.
• After saving, verify the response, re-fetch the instance, and check search or CSW only after the indexing/catalogue gates are known.
• For a declared custom field, use the sparse endpoint only when it is absent from the declared schema: PUT /api/v2/metadata/sparse/<pk>/<key> with {"value":"..."}. Values are strings or null and are limited to 1024 characters; schema-name collisions return HTTP 409.

Metadata, search, and preview handoff

• The default schema is ISO-19115-oriented and includes fields such as title, abstract, date/date type, category, language, license, attribution, regions, keywords, restrictions, temporal extent, maintenance frequency, and spatial representation. Exact type/cardinality comes from the live schema.
• Metadata values also feed generated ISO XML, catalogue AnyText, resource indexes, resource information pages, relations, and discoverability. A thumbnail, map preview, or style is publication/visual context, not a substitute for title, abstract, extent, or keyword metadata.
• For a styled dataset, verify the resource subtype and GeoServer/style gate separately. Keep metadata diagnosis focused on fields and catalogue/index state.

Verify and recover

• Validate local syntax and explicit fields before import. Keep the validator's exit status and messages with the task record.
• Compare schema, submitted instance, saved instance, generated metadata XML, csw_anytext, and search/index results in that order.
• On a failed update, use extraErrors to locate the field/handler; correct the payload and retry without resending unrelated fields.
• On stale discovery, check index freshness and catalogue regeneration rather than declaring the source invalid. Use a deployment-approved regeneration command with a dry run or narrow ID filter before broad changes.
• Report unverified service gates explicitly. See troubleshooting.md for symptom-specific next steps and catalogue-and-search.md for filter semantics.

References and helper

• Metadata reference: handlers, field mappings, localization, schemas, instances, and examples.
• Catalogue and search: CSW backends, format roles, indexes, facets, filters, and preview boundary.
• Troubleshooting: owned failure modes and safe recovery decisions.
• scripts/validate-metadata.py: local-only XML/JSON syntax and caller-supplied required-field checker. It does not contact a catalogue or validate against a remote standards service.