gis-to-db

Use this agent BEFORE running any gis-to-db action skill (inspect, scaffold-service, make-cli, convert, add-module). The agent inspects the user's environment (host Python + GDAL availability, Docker availability), reads the target GIS or CAD file's header to detect format, CRS, layer list, geometry types, and feature count, and — when applicable — verifies the target database is reachable. The agent produces a structured "go / no-go" report that lists every problem the action would otherwise hit at runtime, plus prompts the user about reprojection, layer filters, and source CRS (required for DWG/DXF). Examples: <example> Context: User invokes `/gis-to-db:scaffold-service` and the scaffold-service skill needs to run pre-flight checks before generating files. user: "/gis-to-db:scaffold-service --db-target postgis --name parcels-ingest" assistant: "I'll launch the gis-preflight-validator agent to verify the environment before scaffolding." <Task tool invocation to gis-preflight-validator with mode=scaffold-service> <commentary> Every action skill must run this agent first to fail fast on missing tooling rather than after generating dozens of files. </commentary> </example> <example> Context: User asks to convert a DWG file inline. user: "/gis-to-db:convert ./site-plan.dwg --target postgis" assistant: "Running the gis-preflight-validator first — DWG files require source CRS and layer selection prompts." <Task tool invocation to gis-preflight-validator with mode=convert and file=./site-plan.dwg> <commentary> The validator detects that the DWG has no embedded CRS and 14 layers, then prompts the user for source CRS and which layers to ingest. Without this step, the conversion would silently produce garbage coordinates. </commentary> </example> <example> Context: User wants to add a GIS module to a NestJS+Angular app. user: "/gis-to-db:add-module --host-dir ~/projects/realestate" assistant: "Validating the host application stack and Docker availability before scaffolding." <Task tool invocation to gis-preflight-validator with mode=add-module and host-dir=~/projects/realestate> <commentary> The validator inspects manifests to confirm a supported backend (NestJS) and frontend (Angular) are present, checks for uncommitted changes in the host repo, and ensures Docker is running for the sidecar build step. </commentary> </example>

> **v0.1 status — design doc only.** Pre-built templates land in v0.2 (Slice C). In v0.1, Claude generates the feature module inline from the specs in this SKILL.md. **v0.1 supports only one stack tuple end-to-end: NestJS backend + Angular frontend + PostGIS + Leaflet map preview.** Other backend/frontend/DB/map-library combinations are documented here but require Claude to extrapolate templates — quality varies. For an existing app on a different stack, generate the Python sidecar first via `scaffold-service` and wire the host module manually.

This skill should be used when the user asks to "compare these two GIS datasets", "what changed between this cadastre version and that one", "diff these shapefiles", "find added or removed parcels", "before and after comparison", "what features moved", or invokes `/gis-to-db:analyze-diff`. Compares two GIS layers — typically two versions of the same dataset over time — and reports added features, removed features, geometry changes (movement, area drift), and attribute changes per feature. Matches features by an attribute key (e.g. `parcel_id`) or by geometric proximity when no key is supplied.

This skill should be used when the user asks to "find clusters in this data", "are these points clustered or random", "show me hot spots", "kernel density of features", "spatial clustering with DBSCAN", "find areas with high point density", "nearest neighbor analysis", "spatial pattern detection", or invokes `/gis-to-db:analyze-patterns`. Runs spatial pattern analysis on a GIS point layer: DBSCAN clustering, nearest-neighbor distance distribution, kernel density estimation summary, and convex hulls of detected clusters. Returns labeled features (which cluster each point belongs to) plus a summary report.

This skill should be used when the user asks to "evaluate this location", "is this a good site for X", "site suitability analysis", "why is this location good/bad for an airport / hospital / residential development / factory / solar farm", "score this candidate site", "compare two locations for purpose Y", "site selection criteria", or invokes `/gis-to-db:analyze-site`. The skill scores a candidate point or polygon against a per-purpose ruleset (terrain slope, distance to roads/cities/water, conflict with protected zones, flood risk, demographic context) and produces a transparent markdown report with weighted criteria, scores, and concrete "why" sentences. Rule-based, explainable — no ML, no training data needed.

This skill should be used when the user asks to "show statistics for this GIS layer", "describe the attribute distributions", "summarize this shapefile", "what's the geometry validity of this dataset", "how many duplicates are in this layer", "histogram of parcel areas", "outliers in this data", "data quality report", or invokes `/gis-to-db:analyze-stats`. Produces descriptive statistics for a GIS layer: attribute distributions (numeric histograms, categorical counts), geometry validity report (invalid / self-intersecting / empty), duplicates by attribute or geometry, area / length distributions, feature count breakdown. Pure read-only analysis on a single layer.