Agents & OrchestrationSûr100100/100
Getting Started (TanStack) — Octane Table
Create and render Table v9 with the octane adapter. Route reusable createTableHook
ou envoie-le directement à ton agent.
Installer dans ton projet
$ curl -fsSL https://arboristhree.xyz/cli/install-skill.mjs -o /tmp/arboris-install-skill.mjs && node /tmp/arboris-install-skill.mjs getting-started__tanstack-table__packages-octane-table-skills-b19a3be7a420TanStack
TanStack/table
Contenu à copier
---
name: getting-started
description:
Create and render Table v9 with the octane adapter. Route reusable createTableHook
components, and framework setup; use table-state for reactive ownership.
metadata:
origin: TanStack
type: framework
library: '@tanstack/octane-table'
framework: octane
library_version: '9.3.1'
requires:
- '@tanstack/table-core#core'
sources:
- TanStack/table:docs/framework/octane/quick-start.md
- TanStack/table:examples/octane/basic-use-table
- TanStack/table:packages/octane-table/src/index.ts
- TanStack/table:packages/octane-table/src/useTable.tsrx
- TanStack/table:docs/framework/octane/guide/composable-tables.md
- TanStack/table:docs/framework/octane/guide/table-context.md
- TanStack/table:examples/octane/composable-tables
- TanStack/table:packages/octane-table/src/createTableHook.tsrx
- TanStack/table:packages/octane-table/src/createTableHookContexts.ts
---
Load `intent load @tanstack/table-core#core` first for the headless model, stable inputs, and column inference.
## Setup
```tsrx
import { createRoot, useState } from 'octane'
import {
createColumnHelper,
tableFeatures,
useTable,
} from '@tanstack/octane-table'
type Person = { name: string }
const features = tableFeatures({})
const helper = createColumnHelper<typeof features, Person>()
const columns = helper.columns([helper.accessor('name', { header: 'Name' })])
function PeopleTable() @{
const [data] = useState<Person[]>([{ name: 'Ada' }])
const table = useTable({ features, columns, data })
<table>
<thead>
@for (const group of table.getHeaderGroups(); key group.id) {
<tr>
@for (const header of group.headers; key header.id) {
<th><table.FlexRender header={header} /></th>
}
</tr>
}
</thead>
<tbody>
@for (const row of table.getRowModel().rows; key row.id) {
<tr>
@for (const cell of row.getAllCells(); key cell.id) {
<td><table.FlexRender cell={cell} /></td>
}
</tr>
}
</tbody>
</table>
}
const rootElement = document.getElementById('root')
if (!rootElement) throw new Error('Root element not found')
createRoot(rootElement).render(PeopleTable)
```
## Construction and rendering
This adapter distributes authored TypeScript and TSRX. Configure the consumer with `octane()` in `vite.config.ts` and the TSRX JSX import source `octane`. Author component bodies with `@{ ... }` and keyed `@for` loops for rows, headers, and cells.
Import from `@tanstack/octane-table`. Render `table.FlexRender` as a component so it gets its own Octane scope. It preserves numbers including `0`, render descriptors, and component functions.
Keep features and column definitions at module scope. For inputs that depend on component state, use `useMemo`; pass the current stable data array to `useTable`. Convert non-renderable values with `String(value)` when displaying them as text.
## Read for the task
- When adding or configuring optional features, load `intent load @tanstack/table-core#table-features` and read only references for the requested behavior.
- For state ownership or reactive reads, read [table-state](../table-state/SKILL.md).
- When tables share features, defaults, or reusable UI, read [create-table-hook](references/create-table-hook.md).
## API discovery
Inspect `node_modules/@tanstack/octane-table/src/index.d.ts`, the matching `*.tsrx.d.ts` sidecar, and `src/types.ts`. This package publishes authored source; core APIs are in installed `@tanstack/table-core/dist/`.Colle ce Markdown dans ton agent ou utilise les boutons ci-dessus pour l’écrire dans ton projet.
Ce que fait Getting Started (TanStack) — Octane Table
When adding or configuring optional features, load `intent load @tanstack/table-core#table-features` and read only references for the requested behavior.
For state ownership or reactive reads, read [table-state](../table-state/SKILL.md).
When tables share features, defaults, or reusable UI, read [create-table-hook](references/create-table-hook.md).
Comment utiliser Getting Started (TanStack) — Octane Table
1
Copie le prompt
Un clic copie le prompt packagé (ou l'envoie à ton agent).
2
L'agent installe le skill
Il ajoute le SKILL.md et ses ressources à ton projet.
3
Activation automatique
Le skill s'active dès que le contexte correspond.
Déclencheurs pour Getting Started (TanStack) — Octane Table
Dis simplement à ton agent quelque chose comme :
Applique le skill getting-started__tanstack-table__packages-octane-table-skills-b19a3be7a420 à cette tâche
Utilise getting-started__tanstack-table__packages-octane-table-skills-b19a3be7a420 pour améliorer cette implémentation
Passe en revue ce sujet avec getting-started__tanstack-table__packages-octane-table-skills-b19a3be7a420
Skills liés à Getting Started (TanStack) — Octane Table
Acceleration (Microsoft)
Compare synthetic acceleration options and clearly separate scenario value from forecast commitments.Account Overview (Microsoft)
Create a synthetic Acme account overview and identify evidence that merits seller attention.Acroform Writer (Microsoft)
Use this skill whenever the user wants values written INTO an existing PDF form that lives in the connected SharePoint knowledge source — filling out, completing, populating, or submitting a fillable PDF (application, intake sheet, contract, government form) from data they supply, a spreadsheet row, or the conversation. Triggers include "fill out the job application form", "complete the intake form for [person]", "populate our standard NDA template", and "make this ready to send / non-editable" (flatten). The source form should be found via knowledge search; only ask the user to upload it if knowledge search turns up nothing plausible. different task), and do NOT use it on scanned/image-only PDFs that have no real fillable fields — those need OCR or manual overlay instead, which this skill explicitly detects and reports rather than silently failing.