Claude for QGIS: Scripts, Analysis, and Project Context

Claude can help QGIS users write PyQGIS scripts, explain spatial expressions and functions, troubleshoot layer and analysis problems, and review methodology documentation. There is no Anthropic-built QGIS plugin; you draft in Claude and run the result in QGIS yourself. Tools like fenbs retain source data links, methods decisions, and follow-up tasks so the next analysis session starts with the agreed approach rather than reconstructing it.

Spatial analysis produces outputs that inform real decisions. A script that selects the wrong features, mishandles coordinate reference systems, or applies an incorrect buffer distance produces results that look plausible but are wrong. Verify inputs and check outputs against known reference data at each stage.

What Claude can help with in QGIS

• Writing PyQGIS scripts for the QGIS Python console or as standalone scripts.
• Explaining and debugging QGIS field calculator expressions.
• Describing how to use specific QGIS geoprocessing algorithms and their parameters.
• Reviewing analysis methodology notes for completeness and reproducibility.
• Drafting SQL queries for the QGIS DB Manager or PostGIS layers.
• Explaining coordinate reference system concepts and reprojection decisions.

Writing PyQGIS scripts with Claude

Describe the task in plain terms: what layers you are working with, what the script should do, and what the expected output is. Include the QGIS version. PyQGIS API classes and method signatures change between QGIS versions, so specifying the version helps Claude target the right API.

I am using QGIS 3.36. Write a PyQGIS script that takes the active polygon layer, applies a 500-meter buffer using the project CRS, and saves the result as a new GeoPackage file at a path I supply. The buffer should dissolve overlapping features.

Test every script on a small sample of features before running it on the full dataset. Check that the output CRS is correct, that the attribute table is as expected, and that the geometries look right in the map canvas before proceeding.

Field calculator expressions and spatial SQL

Paste the expression you are trying to write and describe what it should produce. Claude can explain why an expression is returning an unexpected result, suggest an alternative syntax, or draft a PostGIS SQL query for a spatial join or aggregation. For PostGIS queries, specify the PostGIS version and whether you need a specific spatial index behavior.

QGIS expressions and PostGIS SQL differ in syntax and available functions. If you copy a PostGIS example into the QGIS field calculator, it may not work. Tell Claude which environment you are targeting and ask it to use the appropriate function set.

Checking spatial assumptions

Before running an analysis, paste a description of your methodology and ask Claude to identify assumptions that could affect the result. Common ones include CRS mismatch between layers, topology errors in polygon data, null geometries in join fields, and distance calculations that depend on the projection units.

Claude can flag likely problem areas, but it cannot see your data. Check each flagged assumption against the actual layer properties in QGIS. A CRS mismatch that looks fine in the canvas can still produce incorrect distance or area calculations if the analysis algorithm does not reproject on the fly.

Documenting reproducible analysis

Reproducibility requires that someone else — including you, six months later — can recreate the result from the same inputs. Ask Claude to help you draft a methods section that lists each geoprocessing step, the parameters used, the input and output layers, and the software version. Paste the steps you ran and ask Claude to turn them into structured documentation.

Record which version of each source dataset you used and where it came from. Public datasets are updated and corrected over time; the version used in an analysis should be identifiable if someone questions the result later.

How to do this with fenbs

1. Create a free account and add a project for this spatial analysis.
2. Add each source dataset as a link entry: the dataset name, the source agency URL, the download date, and the CRS.
3. Record methods decisions — such as which buffer distance was agreed with the client and why — as Decisions and rules.
4. Track open analysis tasks and follow-up checks as tasks on the board.
5. Before each session, review the source links and decisions to confirm the current agreed methodology.

Automating QGIS workflows

For processing workflows you run repeatedly — such as downloading updated datasets, reprojecting them, and running a standard analysis sequence — ask Claude to draft a standalone Python script that uses QGIS's headless processing mode. Describe each step in the sequence, the input file format and CRS, and where the output should be saved. A well-structured script is faster and more reproducible than running the same geoprocessing steps manually each time.

Test the script on a small, known dataset before pointing it at live data. Check that the output matches a manually produced result for the same input. Automated workflows can mask errors that a manual step would have made visible.

What Claude cannot do in QGIS

Claude cannot open GeoPackage, Shapefile, or raster files, read layer properties, or run geoprocessing algorithms. It works from text you paste. For spatial questions that require seeing the data — such as why a join is returning null values — you need to inspect the actual layers in QGIS. See what is Claude AI and what is Claude Code for context on what Claude can and cannot access.