Repository navigation
Status and open work
This page comes from
site/content/project/status.mdin the repository and is regenerated on every merge tomain. To correct it, edit that file; changes made here are overwritten.
Cartogen AI is a release-candidate series for 1.16.0 (current build 1.18.5); the last stable release is v1.15.6. Each release states whether it was checked by CI only or also by hand.
| Checked by | State |
|---|---|
| Offline test suite (6,993 test functions) | Passing on every merged change, on Python 3.10 and 3.11 and Windows, plus ruff and mypy. |
| QGIS 4.2.2 live tests in CI | Passing, including the widget, network, chunking, raster and PostGIS tests. |
| A person in desktop QGIS | Partial. The last full hand run was rc22 (2026-10-09). The rc37 pack was run on 2026-10-13: of 23 rows, 17 passed, 2 were inconclusive, 2 were not recorded and 2 optional rows were skipped. Rows for rc23 to rc36 and most of the 179 older rows have not been hand-run. |
Found in the rc37 run and fixed in the following candidates:
-
rc38:
run_allowlisted_processing_algorithmhad failed on every model call since rc24 (a decorator had been left on a helper function), and a false "step did not run" note appeared after incident searches. - rc39: the follow-up that told the model to "generate a representative sample dataset" when no layer existed was removed, after the model put four invented incidents on the map.
These fixes are CI-checked; they have not been re-run by hand.
- Hand-test the current candidate: install gate; the incident prompt must add no invented layer; the Processing-tool follow-up; Stop heard during a real long job (still unproven); then the 16 categories and the older rows.
- Upgrade gate: a clean-profile install and an in-place upgrade from v1.15.6.
- plugins.qgis.org: upload and pass its security scan.
-
Open issues (46 at the last count): each closes only on a hand-verified pass of its own row, with evidence. Six items sit inside closed issues and need a reopen-or-file decision (DEM vertical unit assumed metres; score writes in
analysis_tools; raster-unit validation in zonal tools; reprojecting points into the admin CRS; running tasks not invalidated at unload; the wheel not built in a clean environment). - JIAF validation blockers: a real country's Flags export and a real filled Annex 4 file are needed to close two blockers, or stable ships with them still shown in every result.
-
Version and notes: switch to
1.16.0, remove the pre-release wording, run the release workflow, promote to "Latest".
-
Invented records by another route. The wording that invited fabrication is gone, but nothing stops the model adding made-up points with a tool such as
add_point_layer. - Stop on a long job has been tested with fakes and live tests, not on a large real project; the speed-up from the responsiveness work has not been measured.
-
Python Processing algorithms (
gdal:*,qgis:*) stay on the main thread by decision; a CI investigation could not reproduce the old crash on a worker thread, but nothing proves it is safe. - A route request used 9 model calls for 5 stops, over the 4-call guideline.
- Credentialed integrations (Kobo/ODK, ACLED, GloFAS/CHIRPS, Logistics Cluster, LandScan) are deferred; ACLED and Liveuamap are not searched by the incident tool.
- The hosted gateway and the "Cartogen API" provider are a stub. No hosted service exists.
- GDPR: the repository carries a code-level review and a DPIA screening worksheet; there is no legal sign-off and no built-in redaction. See What leaves your machine.
Statuses are written in the code and docs as shipped, roadmap or unverified; frozen historical documents are never edited after the fact (a newer dated document supersedes them); and a judgement call is flagged with a recommendation, not decided silently. See Contributing.
Cartogen AI 1.18.5 · generated from the repository at commit 26536e9f75 on 2026-11-12 · this wiki is rebuilt on every merge to main; edits made in the browser are overwritten, so correct the source files instead.
Get started
How it works
- Architecture overview
- The agent loop
- Threads, tasks and Stop
- Tools and routing
- Providers and networking
- Memory, state and results
- Guards and honesty checks
- The dock and dialogs
- Processing provider
Tools and data
- Tool reference
- AI Imagery Feature Extraction
- Data Analysis & Prediction
- Database & Workflows
- Export & Reporting
- Humanitarian Data (HDX / OSM / geoBoundaries)
- Humanitarian Logistics
- Monitoring & Scheduling
- Print Layouts
- Project Management
- Raster
- Reporting & Document Analysis
- Satellite Imagery & Vision
- Styling & Labeling
- System, Search & Scripting
- Task & Memory Management
- Vector & Geoprocessing
- More tools, by module
- Task register
- External data sources
Humanitarian GIS
Security and privacy
Develop
Project