Agent skill
crop-data
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/crop-data
SKILL.md
Crop Data Context
This skill provides essential context for working with the crop planning data model.
CRITICAL: Data Analysis Caveats
The extracted analysis files have known gaps:
| File | Issue |
|---|---|
column-analysis.json |
Misclassifies array formulas as "static" or "mixed" |
formula-analysis.json |
Missing ~20+ formulas using SWITCH, complex XLOOKUP |
Always verify against the actual workbook using:
python scripts/inspect-column.py "STH" # Inspect specific column
python scripts/inspect-column.py --list # List all columns
python scripts/inspect-column.py "DTM" --sheet "Bed Plan"
Authoritative reference: vba/workbook-structure.md has table schemas extracted directly from Excel.
Key Timing Formulas (FROM WORKBOOK)
STH (Seed To Harvest) - NOT static, despite what analysis files say:
SWITCH(Normal Method,
"DS": Days to Germination + DTM + IF(Days in Cells > 0, 15, 0)
"TP": DTM + Days in Cells + IF(Days in Cells <= 0, 20, 0)
"X": DTM
)
DTM - Calculated from range:
=FLOOR.MATH(AVERAGE(DTM Lower, DTM Upper))
Days in Cells - Sum of tray stages:
=SUM(Tray 1 Days, Tray 2 Days, Tray 3 Days)
Target dates:
Target Sewing Date = LastFrostDate - 5 * Sewing Rel Last FrostTarget Field Date = Target Sewing Date + Days in CellsTarget Harvest Data = Target Sewing Date + STHTarget End of Harvest = Target Harvest Data + Harvest window
Entity Model
GLOBAL CONSTANTS (farm settings: LastFrostDate, BedLength, LaborRate)
│
▼
CROP (~100 unique) ─────────────────────────────────────────────────┐
│ name, category, deprecated │
│ Same for all configurations of a crop │
│ │
│ one crop → many configurations │
▼ │
PLANTING CONFIGURATION (340 total) │
│ structure (Field/Hoop/Greenhouse) │
│ method (DS/TP), season (Sp/Su/Fa/Wi/Ow) │
│ product type, DTM, spacing, rows │
│ Identifier: "Arugula - Baby Leaf | Field DS Sp" │
│ │ │
│ one config → many plantings │ references │
▼ ▼ │
PLANTING (per-plan instance) PRODUCT (processing info) ◄──┘
│ bed, bedsCount, startDate │ Unit, Price, Yield
│ per-planting overrides │ Wash type, labor times
│ actual vs planned dates (future) │ Holding period
Data Source Files
| File | Size | Contents | Notes |
|---|---|---|---|
src/data/crops.json |
1.7 MB | 340 configs × ~155 fields | Fat export, includes calculated |
src/data/bed-plan.json |
170 KB | ~138 plantings | Current plan assignments |
src/data/products.json |
213 KB | Product processing data | Labor times, packaging |
column-analysis.json |
46 KB | Field classification | Has gaps - verify! |
formula-analysis.json |
29 KB | Formula DAG | Missing array formulas! |
vba/workbook-structure.md |
- | Table schemas | Authoritative reference |
Planting Override Fields
These bed-plan fields modify catalog values per-planting:
| Override Field | Modifies | Purpose |
|---|---|---|
additionalDaysInField |
STH/DTM | Weather, conditions adjustment |
additionalDaysOfHarvest |
Harvest window | Extended/shortened season |
additionalDaysInCells |
Days in Cells | Greenhouse timing adjustment |
Computation pattern:
const effectiveDTM = catalog.STH + planting.additionalDaysInField;
const effectiveHarvestWindow = catalog.harvestWindow + planting.additionalDaysOfHarvest;
Timeline Date Computation
The timeline uses these inputs to compute dates:
From Catalog (crops.json):
STH- Total days from seeding to first harvestHarvest window- Days of harvest periodDays in Cells- Greenhouse time (0 = direct seed)
From Planting (bed-plan.json):
fixedFieldStartDate- The target transplant/direct-seed dateadditionalDays*- Per-planting adjustments
Computed:
startDate = fixedFieldStartDate (the TP/DS date)
harvestStartDate = startDate + (STH - daysInCells) = startDate + DTM
endDate = harvestStartDate + harvestWindow
Finding Things
Run from crop-api/ directory. Use patterns (not file paths) to locate code:
# Entity types (new canonical location)
node scripts/ast-query.js "Plan" # What does Plan contain?
node scripts/ast-query.js "Planting" # What does Planting contain?
node scripts/ast-query.js "CropConfig" # What does CropConfig contain?
# Find type definitions
grep -r "^export interface" src/lib/entities/
# Find calculation functions
grep -r "^export function calculate" src/lib
# Find what uses entities
grep -r "from.*entities" src/lib --include="*.ts"
# Find bed length logic
grep -r "lengthFt\|ROW_LENGTHS" src/lib
# Find plan store actions
grep -r "^ [a-z]*:" src/lib/plan-store.ts | head -30
Data Extraction
Re-extract from Excel when workbook changes (run from crop-api/):
python scripts/extract-products.py
Testing & Verification
Run from crop-api/:
npx tsx scripts/test-entities.ts # Entity validation
npx tsx scripts/test-slim-planting.ts # Date computation parity
npx tsx scripts/test-catalog-parity.ts # Catalog lookup parity
npx tsx scripts/test-crop-calculations.ts # Crop calc vs Excel
# Inspect actual Excel formulas
python scripts/inspect-column.py "STH"
python scripts/inspect-column.py --list
Architecture
New canonical types live in lib/entities/:
Bed- bed withlengthFt(F/J=20, A-E/G-I/U=50, X=80)Planting- one per planting, referencesconfigIdCropConfig- planting configuration with calculationsPlan- self-contained: ownsbeds,cropCatalog,plantings
Key principle: Plans own their data. No external references.
// What we store
interface Plan {
beds: Record<string, Bed>; // Plan's bed layout
cropCatalog: Record<string, CropConfig>; // Plan's crop configs
plantings: Planting[]; // One per planting decision
}
// Computed at render time
const timing = calculateCropTiming(planting, config);
const bedSpan = calculateBedSpan(planting.bedFeet, planting.startBed, plan.beds);
See .claude/plans/state-migration.md for full migration plan.
Didn't find tool you were looking for?