deterministic-scripts
Skill110 starsChanged 3 months agoMITView on GitHub
Deterministic script execution skill for HASTE. Execute scripts instead of free-form LLM behavior for consistent, repeatable operations. Use when: 'run preprocessing', 'convert format', 'parse metadata', 'generate tiles', 'build wheel', 'deploy functions'. Prevents hallucinated scripts.
- Installs packages
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- api-designSkill
- copilot-cli-modesSkill
- debug-test-failuresSkill
- dependency-updateSkill
- imagery-provider-adaptationSkill
- release-notesSkill
- security-analysisSkill
- validation-diagnosticsSkill
The file
Files
Deterministic Script Execution
Overview
HASTE has established scripts and commands for common operations. Agents must use these exact commands instead of generating free-form alternatives. This prevents hallucinated scripts, ensures consistency, and makes operations repeatable.
Key Concepts
Why Deterministic Scripts Matter
- LLMs can "helpfully" generate plausible-looking but incorrect commands
- HASTE has specific build tooling (hatch, conda) that must be used correctly
- Azure Functions deployment has specific prerequisites and ordering
- Geospatial processing requires exact GDAL/rasterio invocations
Patterns & Techniques
Build & Package Operations
| Operation | Exact Command | Notes |
| Build core wheel | cd hastelib && hatch build -t wheel | Auto-increments version, copies to func apps |
| Run Python tests | cd hastelib && hatch run test:pytest | Uses conda env with GDAL |
| Build UI | cd ui && npm run build | Production build via Vite |
| Lint UI | cd ui && npm run lint | ESLint with React rules |
| Install UI deps | cd ui && npm install | Uses package-lock.json |
| Create conda env | conda env create -f env.yml | Full env with GDAL and dependencies |
| Update conda env | conda env update -f env.yml | Preserves existing packages |
| Install hastelib editable | pip install -e hastelib/ | For local development hot-reload |
Local Development
| Operation | Exact Command | Notes |
| Start API locally | cd api/hastefuncapi && func host start | Requires .venv or conda env |
| Start UI locally | cd ui && swa start --app-devserver-url http://localhost:5173 --run 'npm run dev' | SWA CLI with Vite dev server |
| Start Azurite | azurite --silent --location ./data --debug ./data/debug.log | Local Azure Storage emulator |
| Start Docker stack | docker-compose -f docker/docker-compose.yml up | Full local stack |
Deployment
| Operation | Exact Command | Notes |
| Deploy Azure Function | func azure functionapp publish <NAME> --subscription <SUB> --tenant <TENANT> | After hatch build |
| Deploy SWA | cd ui && swa deploy --app-location ./dist --app-name <NAME> | After npm run build |
| Build Docker images | ./build_and_push_images.sh | Builds training + imagery prep images |
Imagery Processing
| Operation | Approach | Notes |
| COG generation | Use rasterio with COG driver profile | Never use raw GDAL CLI unless wrapping in Python |
| Tile generation | Through ImageryPreProcessor | Not manual gdal2tiles |
| Reprojection | rasterio.warp.reproject() | Always specify src_crs and dst_crs |
| Format conversion | Through provider-specific adapter | Not generic gdal_translate |
Decision Framework
| Situation | Do This | NOT This |
| Need to run tests | hatch run test:pytest | pytest (wrong env) |
| Need to build wheel | hatch build -t wheel | python setup.py bdist_wheel |
| Need to start API | func host start | python function_app.py |
| Need Azure storage locally | azurite | Custom mock storage |
| Need to process imagery | Use ImageryPreProcessor | Write new GDAL script |
| Need to deploy | func azure functionapp publish | Manual zip deployment |
Common Pitfalls
- Inventing new build commands — Use the established commands above
- Running pytest directly — Use
hatch run test:pytestto get the correct conda env - Starting the UI with
npm start— HASTE usesswa startwith Vite dev server - Using
python setup.py— HASTE uses hatch/hatchling, not setuptools - Generating GDAL scripts from scratch — Use existing processor methods
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.