useActivityLog

Writes audit entries to the backend. useCrud already calls it on every create, update and delete.

Import

import { activityEntityLabel, useActivityLog } from '@shared/composables/useActivityLog'

Signature

function useActivityLog(): {
  log: (entry: Omit<ActivityEntry, 'id' | 'timestamp'>) => Promise<void>
}

function activityEntityLabel(entity: string): string

Returns

Member	Type	Description
log	(entry) => Promise<void>	POSTs to /activity-log. The id and timestamp are the server's.

ActivityEntry

Key	Type	Description
id	string	Assigned by the backend.
action	'create' | 'update' | 'delete' | string	A free string, so an app can log its own verbs.
entity	string	The resource. A stable identifier, not a label.
entityId	string | number	Which row.
details	Record<string, unknown>	Whatever is worth keeping — usually the payload.
userId	string	Who did it.
timestamp	Date	Assigned by the backend.

activityEntityLabel

Turns an entity string into something readable. It looks for activity-log.entities.{entity} in lower case and returns the translation if it exists, or the entity as logged if it does not.

That fallback matters because entity is a free string — a module name, a table name, whatever the app passed. Only the ones with a key translate; the rest render exactly as they were recorded.

Example

<script setup lang="ts">
import { activityEntityLabel, useActivityLog } from '@shared/composables/useActivityLog'

const { log } = useActivityLog()

async function approve(id: number) {
  await fetch(`/api/expenses/${id}/approve`, { method: 'POST' })
  await log({
    action: 'approve',
    entity: 'expense',
    entityId: id,
    details: { approvedBy: 'finance' },
  })
}
</script>

<template>
  <div>
    <button type="button" @click="approve(41)">
      Approve
    </button>
    <p>{{ activityEntityLabel('expense') }}</p>
  </div>
</template>

Notes

• entity must not change with the language. It is data in the audit trail, so it is 'warehouse' — never "Warehouse", never a translated word. activityEntityLabel is what turns it into something to read, at the point it is displayed.
• log is a plain POST and it rejects on failure. useCrud calls it fire-and-forget, so a failed audit never breaks the operation that triggered it. Do the same if the entry is not more important than the action.
• The activity-log page reads entries back through its own useCrud instance, and sets activityLog: false there so reading the log does not write to it.
• The whole feature sits behind config.features.activityLog.

Related

• useCrud — the automatic caller.
• Auth and permissions — who is recorded.