This skill builds on core. Read it first for the headless model and stable inputs.

Setup

import {
  rowAggregationFeature,
  aggregationFn_sum,
  columnGroupingFeature,
  createFilteredRowModel,
  createSortedRowModel,
  columnFilteringFeature,
  filterFn_includesString,
  rowSortingFeature,
  sortFn_alphanumeric,
  tableFeatures,
} from '@tanstack/table-core'

export const features = tableFeatures({
  columnFilteringFeature,
  filteredRowModel: createFilteredRowModel(),
  filterFns: { includesString: filterFn_includesString },
  rowAggregationFeature,
  columnGroupingFeature,
  aggregationFns: { sum: aggregationFn_sum },
  rowSortingFeature,
  sortedRowModel: createSortedRowModel(),
  sortFns: { alphanumeric: sortFn_alphanumeric },
})

Core Patterns

Register feature before its dependent slot

const features = tableFeatures({
  rowSortingFeature,
  sortedRowModel: createSortedRowModel(),
})

tableFeatures checks slot prerequisites and its inferred type gates APIs throughout the table.

Register named function slots with their features

const features = tableFeatures({
  columnFilteringFeature,
  filterFns: { includesString: filterFn_includesString },
  rowSortingFeature,
  sortFns: { alphanumeric: sortFn_alphanumeric },
  rowAggregationFeature,
  columnGroupingFeature,
  aggregationFns: { sum: aggregationFn_sum },
})

filterFns, sortFns, and aggregationFns are feature slots, not table options. They respectively require columnFilteringFeature, rowSortingFeature, and rowAggregationFeature. Import individual built-ins (filterFn_*, sortFn_*, aggregationFn_*) and register them under their conventional keys; the full registry objects (filterFns, sortFns, aggregationFns exports) still work but bundle every built-in. A registered key can be used as a typed string name, and 'auto' resolves only registered functions; pass a function directly when no registry name is needed.

Prefer explicit features

const features = tableFeatures({ columnFilteringFeature })

Use stockFeatures only for deliberate kitchen-sink or temporary migration behavior.

Common Mistakes

[CRITICAL] Calling an unregistered feature API

Wrong:

const features = tableFeatures({})
table.setSorting([{ id: 'name', desc: false }])

Correct:

const features = tableFeatures({ rowSortingFeature })

Optional feature state and APIs are installed only when their feature is registered.

Source: packages/table-core/src/core/table/constructTable.ts

[HIGH] Omitting a slot prerequisite

Wrong:

const features = tableFeatures({ sortedRowModel: createSortedRowModel() })

Correct:

const features = tableFeatures({
  rowSortingFeature,
  sortedRowModel: createSortedRowModel(),
})

The sorted model slot requires rowSortingFeature; the same rule applies to every mapped slot.

Source: packages/table-core/src/types/TableFeatures.ts#FeatureSlotPrereqs

[MEDIUM] Shipping all features by default

Wrong:

const features = stockFeatures

Correct:

const features = tableFeatures({
  rowSortingFeature,
  sortedRowModel: createSortedRowModel(),
})

stockFeatures registers every stock plugin and processing slot, defeating v9's normal tree-shaking strategy.

Source: packages/table-core/src/features/stockFeatures.ts

API Discovery

Inspect node_modules/@tanstack/table-core/dist/types/TableFeatures.d.ts for current slots and FeatureSlotPrereqs, and dist/features/stockFeatures.d.ts for the stock inventory.