Feature architecture

Read core first for the headless model and stable inputs.

Register the behavior the table uses

<!-- skill-snippet:check -->

import {
  createSortedRowModel,
  rowSortingFeature,
  sortFn_alphanumeric,
  tableFeatures,
} from '@tanstack/table-core'

export const features = tableFeatures({
  rowSortingFeature,
  sortedRowModel: createSortedRowModel(),
  sortFns: { alphanumeric: sortFn_alphanumeric },
})

Pass the stable features object to the installed adapter constructor. The concrete registry determines optional APIs, options, state slices, and their types. The core row model is automatic; a custom core model uses coreRowModel.

Registration rules

• Register each prerequisite feature before its dependent slot in the same tableFeatures call. A processing factory alone cannot install that feature's APIs or state.
• Row-model factories use create*RowModel() feature slots. Their names and prerequisites are in the feature reference. They take no function-registry arguments.
• filterFns, sortFns, and aggregationFns are feature slots, not table options. They require columnFilteringFeature, rowSortingFeature, and rowAggregationFeature, respectively.
• Import individual built-ins such as sortFn_alphanumeric and register their conventional keys. Those keys become typed string names; 'auto' can resolve only registered functions. Pass a function directly when a column needs no named registry entry. Full registry exports bundle every built-in.
• columnResizingFeature requires columnSizingFeature; globalFilteringFeature requires columnFilteringFeature. Check installed FeatureSlotPrereqs for other dependencies.
• Use explicit features for normal construction. stockFeatures is a deliberate all-features convenience or temporary migration aid, with the corresponding bundle cost.

A missing API can mean missing registration. Check the concrete feature object before casting, recreating an API, or assuming v9 removed it.

Select references by the requested change

Read only references needed for the requested behavior and its dependencies, including features being added. An unrelated feature already registered on the table does not make its reference necessary.

Task involves	Read
Manual modes, server data, mixed row-processing stages, or dataset scope	Client/server ownership
Totals, multiple aggregations, grouped aggregate values, or custom aggregate definitions	Aggregation
Facet option counts, numeric ranges, or incomplete server facets	Column faceting
Per-column filters, filter functions, metadata, or nested-row filtering	Column filtering
Group rows, placeholders, or grouping with expansion/pagination	Grouping
Drag ordering or leaf-column order differing from state	Column ordering
Sticky column regions, logical start/end, or pinning gaps	Column pinning
Drag resize handles, gesture events, or resize performance	Column resizing
Numeric widths, min/max limits, or model/CSS size mismatch	Column sizing
Hidden columns, visibility-aware rendering, or hiding controls	Column visibility
A search across columns or global-filter eligibility	Global filtering
Hierarchical subrows, detail panels, or expansion/page interaction	Expanding
Page slicing, counts, navigation, or page-index resets	Pagination
Top/bottom pinned rows or their visibility outside the current page	Row pinning
Rectangular cell selection, include/exclude ranges, or drag outlines	Cell selection
Merged body cells, covered cells, or spans changing with row order	Cell spanning
Row checkboxes, select-all, Shift ranges, or IDs across pages	Row selection
Sorting, comparators, undefined values, or sort interaction cycles	Sorting

For controlled state or reset behavior, read shared state and load the installed adapter's state skill for reactive wiring.

Installed API discovery

Inspect node_modules/@tanstack/table-core/dist/types/TableFeatures.d.ts for slots and FeatureSlotPrereqs. Follow dist/features/<feature>/ for exact state, option, and instance declarations. Each reference names its feature directory.