The interface is localized. Skill descriptions and documents are machine-translated by AI into your language; code, commands, paths and technical terms stay in the original, and the author's document governs.
Explore skills
DevOps · SKILL.md

portaljs-add-geo

portaljs-add-geo

Auto-ingest a geospatial file (GeoJSON, Shapefile, GeoPackage, KML/KMZ, FlatGeobuf, CSV-with-geometry) into a PortalJS portal on the user's own machine, with no server. Normalizes CRS to EPSG:4326, derives a PMTiles render tier and a GeoParquet query tier, pushes all three artifacts to Cloudflare R2 via Git LFS, and appends one dual-tier datasets.json entry…

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill documentation

GitHub

name: portaljs-add-geo description: Auto-ingest a geospatial file (GeoJSON, Shapefile, GeoPackage, KML/KMZ, FlatGeobuf, CSV-with-geometry) into a PortalJS portal on the user's own machine, with no server. Normalizes CRS to EPSG:4326, derives a PMTiles render tier and a GeoParquet query tier, pushes all three artifacts to Cloudflare R2 via Git LFS, and appends one dual-tier datasets.json entry the showcase auto-renders. Use when the source is a vector geo format that needs a map or spatial-query view. allowed-tools: Read, Write, Edit, Bash(ogr2ogr:), Bash(ogrinfo:), Bash(tippecanoe:), Bash(duckdb:), Bash(git:), Bash(curl:), Bash(npx:), Bash(node:), Bash(mkdir:), Bash(cp:), Bash(wc:), Bash(command:), WebFetch version: 1.0.0 author: Datopian hello@datopian.com license: MIT compatibility: Claude Code with PortalJS portals (Next.js 14, React 18/19, Node 18+). Requires native GDAL, tippecanoe, and duckdb on the user's machine (macOS or Linux/WSL). Runs from any project via the plugin, a personal ~/.claude/commands install, or a portaljs clone. tags:

  • portaljs
  • data-portal
  • geospatial
  • pmtiles
  • geoparquet
  • duckdb

PortalJS — Add Geo

Overview

Turn one geospatial upload into a dual-tier PortalJS dataset entirely on the user's machine — no server-side container compute, keeping the zero-backend bet end-to-end. From a single source the skill derives two serverless tiers plus the preserved original:

  • PMTiles (render tier) — a vector-tile archive <MapPreview> renders with MapLibre GL over HTTP range requests; any dataset size pans and zooms with no tile server.
  • GeoParquet (query tier) — a GeoParquet 1.1 file (covering bbox column, Hilbert-sorted) <GeoQuery> runs spatial SQL over in place via DuckDB-Wasm.
  • Original — the untouched upload in its native CRS, kept downloadable.

All three land on Cloudflare R2 (Git LFS → Giftless), and the skill appends one datasets.json entry whose resources[] match the shape the showcase auto-renders (the @reference/world-boundaries demo, PR #1647): pmtiles → <MapPreview>, geoparquet → <GeoQuery>, original → download. No page edits are needed. It automates the manual tippecanoe/duckdb recipes in the template README.

Prerequisites

  • A scaffolded portaljs-catalog portal whose template ships components/MapPreview.tsx and components/GeoQuery.tsx (PR #1647 or later).
  • Native CLIs on the user's machine: GDAL (ogr2ogr, ogrinfo), tippecanoe, and duckdb (with the spatial extension). macOS: brew install gdal tippecanoe duckdb; Debian/Ubuntu: apt-get install gdal-bin duckdb + tippecanoe (apt or build from source); Windows: use WSL. The skill hard-stops with the install hint if any is missing.
  • Arc credentials for the Git-LFS → R2 push (the same token portaljs-deploy resolves), or an OSS self-hosted Giftless.

Instructions

The canonical, full step-by-step workflow is .claude/commands/portaljs-add-geo.md — the single source of truth. Read and follow it when executing. Summary:

  1. Gather input — source (geo file path or URL), portal directory, namespace. Interview if missing; never dead-end.
  2. Detect the native tools (ogr2ogr, tippecanoe, duckdb + spatial). Any missing → print the per-OS install and stop.
  3. Validate the portal directory and confirm the geo showcase components exist.
  4. Fetch (if URL) and detect the input format; reject raster and bare .shp.
  5. Preserve the original, then normalize to EPSG:4326 GeoJSON with ogr2ogr -t_srs EPSG:4326.
  6. Apply the size escape hatch (measure normalized-GeoJSON bytes): warn ≥ 500 MB; skip PMTiles and emit GeoParquet-only > 2 GB (documented, never silent).
  7. Derive PMTiles with tippecanoe -zg --drop-densest-as-needed.
  8. Derive GeoParquet with duckdb — covering bbox STRUCT + ST_Hilbert order; verify a range read.
  9. Track all three files with Git LFS, mint an Arc JWT, push to R2, prune, and build the absolute data.portaljs.com URLs.
  10. Append one dual-tier datasets.json entry (pmtiles + geoparquet + original resources).
  11. Verify: npx tsc --noEmit and a Range: 206 check on the R2 URL.
  12. Report the tiers, manifest entry, and showcase route.

Output

  • Created: data/<slug>.pmtiles, data/<slug>.parquet, data/<slug>.<origext> (all LFS-tracked → R2); data/<slug>.4326.geojson (intermediate; can be pruned).
  • Modified: datasets.json (one dual-tier entry appended); .gitattributes (LFS tracking).
  • Verified: npx tsc --noEmit passes; the R2 URLs answer HTTP 206 range reads.
  • Result: /@<namespace>/<slug> renders the PMTiles map and the GeoParquet spatial-query panel together, with no page edits.

Error Handling

SymptomCauseFix
MISSING_TOOLSogr2ogr/tippecanoe/duckdb (or duckdb spatial) absentPrint the per-OS install line and stop; re-run after installing.
UNSUPPORTED_FORMATRaster .tif/.tiff (COG tier, later phase)Use portaljs-add-dataset for tabular; await the COG phase.
SHAPEFILE_NOT_ZIPPEDBare .shp with no siblingsZip the whole .shp/.dbf/.shx/.prj set and pass the .zip.
MULTI_LAYERGPKG/KML with more than one layerList layers and ask which to ingest; pass the layer name.
REPROJECTION_FAILEDogr2ogr error, often a missing .prjShow stderr; re-run with -s_srs EPSG:<code>.
EMPTY_OUTPUTNormalized GeoJSON has 0 featuresCheck the geometry column / CRS of the source.
TILING_FAILED / GEOPARQUET_FAILEDtippecanoe or duckdb non-zeroShow stderr; for very large inputs see the size escape hatch.
LFS_PUSH_FAILEDMissing/expired Arc token or unset lfs.urlRe-mint the JWT (see portaljs-deploy); confirm git config lfs.url.

Examples

Example 1 — Ingest a GeoJSON into both tiers

/portaljs-add-geo source=./data/rivers.geojson name="World Rivers" namespace=reference

Example 2 — Zipped Shapefile from a URL

/portaljs-add-geo source=https://example.com/admin-boundaries.zip slug=admin-boundaries

Example 3 — GeoPackage with an explicit layer and namespace

/portaljs-add-geo source=./data/census.gpkg slug=census-tracts namespace=statistics

Resources

Saved skills