44 lines
1.4 KiB
Markdown
44 lines
1.4 KiB
Markdown
# M4 Autonomous Build Readiness
|
|
|
|
This milestone converts GeoIntel from an implementation-ready blueprint into an autonomous-build package for Codex.
|
|
|
|
## Goal
|
|
Codex must be able to start from the repository root, read the contracts, build the foundation, and keep moving through backend, frontend, GIS processing, AI pipeline stubs, QA/QC, tests and documentation without making new product or architecture decisions.
|
|
|
|
## M4 Adds
|
|
- Sprint board with sequential implementation lanes.
|
|
- Module-level build contracts.
|
|
- Acceptance-test catalog per module.
|
|
- Frontend state and page contracts.
|
|
- Backend service IO contracts.
|
|
- Example API responses and error shapes.
|
|
- Job lifecycle contract.
|
|
- Model registry seed contract.
|
|
- Demo fixture manifest.
|
|
- Codex prompts per autonomous build pass.
|
|
- Final readiness checklist.
|
|
|
|
## Golden Rule
|
|
When code conflicts with documentation, documentation wins unless an explicit ADR is added.
|
|
|
|
## Build Priority
|
|
1. Running local stack.
|
|
2. Database models and migrations.
|
|
3. API contracts.
|
|
4. Dataset manager.
|
|
5. Map workspace.
|
|
6. Raster/vector metadata flows.
|
|
7. Analysis run lifecycle.
|
|
8. Detection/segmentation job shells with deterministic demo mode.
|
|
9. QA/QC engine.
|
|
10. Exports.
|
|
11. Polish, tests and documentation.
|
|
|
|
## Do Not Build Yet
|
|
- Full LiDAR workbench.
|
|
- Full training studio.
|
|
- QGIS plugin.
|
|
- Cloud deployment.
|
|
- Real production-grade model training.
|
|
- Complex user auth.
|