## Retrieve

**get** `/v1/enrichment-status`

Returns a tenant's enrichment progress across the record-level enrichments
(translation, sentiment, emotions). For each, `enabled` reports whether the enrichment
is active for the tenant (deployment-configured and switched on / with a resolvable
target language), and `eligible`/`done` are directory-level counts of feedback records
that qualify and that have been enriched — the UI derives "in progress" as
`eligible - done`, of which `failed` is retryable and `failed_terminal` never will be.
When an enrichment is not enabled its counts are zero and
`disabled_reason` names the gate that closed it, so the UI can explain *why* rather than
silently hiding the enrichment. `as_of` is when the counts were computed, which lets a
polling client derive throughput and an ETA without the Hub computing either. The response
contains counts only (no record identifiers or content).

### Query Parameters

- `tenant_id: string`

  Tenant whose enrichment status should be returned.

### Returns

- `as_of: string`

  When the counts were computed (UTC), not when the response was serialized. The
  endpoint is polled, so two responses are enough to derive throughput and an ETA
  client-side from the change in `done` over the change in `as_of` — the Hub computes
  neither.

- `emotions: TypeStatus`

  One enrichment's progress for a tenant. When `enabled` is false, `eligible` and `done` are zero.

  - `done: number`

    Eligible records that have been enriched.

  - `eligible: number`

    Feedback records that qualify for this enrichment.

  - `enabled: boolean`

    Whether the enrichment is active for the tenant (deployment-configured and switched on / with a resolvable target language).

  - `failed: number`

    Eligible records whose last enrichment attempt gave up but which a retry could still
    rescue — a provider outage, a timeout. Counted only while the record is still
    un-enriched, so a later success removes it without any cleanup.

  - `failed_terminal: number`

    Eligible records the provider will never accept, because the outcome is a property of
    the record's own text — a content-policy block, a refusal, an input past the model's
    limit. Retrying these cannot help; they resolve only if the text changes.

  - `disabled_reason: optional "not_configured" or "switched_off" or "no_target_language"`

    Which gate switched the enrichment off. Present exactly when `enabled` is false, and
    absent otherwise. `not_configured` — the deployment has no provider/model for this
    enrichment, so it is off for every tenant and only an operator can fix it.
    `switched_off` — the tenant turned it off (sentiment and emotions only).
    `no_target_language` — translation is configured but neither the tenant's
    `target_language` nor the deployment default resolves (translation only; it has no
    on/off switch, so an absent target is its off state).

    - `"not_configured"`

    - `"switched_off"`

    - `"no_target_language"`

- `sentiment: TypeStatus`

  One enrichment's progress for a tenant. When `enabled` is false, `eligible` and `done` are zero.

  - `done: number`

    Eligible records that have been enriched.

  - `eligible: number`

    Feedback records that qualify for this enrichment.

  - `enabled: boolean`

    Whether the enrichment is active for the tenant (deployment-configured and switched on / with a resolvable target language).

  - `failed: number`

    Eligible records whose last enrichment attempt gave up but which a retry could still
    rescue — a provider outage, a timeout. Counted only while the record is still
    un-enriched, so a later success removes it without any cleanup.

  - `failed_terminal: number`

    Eligible records the provider will never accept, because the outcome is a property of
    the record's own text — a content-policy block, a refusal, an input past the model's
    limit. Retrying these cannot help; they resolve only if the text changes.

  - `disabled_reason: optional "not_configured" or "switched_off" or "no_target_language"`

    Which gate switched the enrichment off. Present exactly when `enabled` is false, and
    absent otherwise. `not_configured` — the deployment has no provider/model for this
    enrichment, so it is off for every tenant and only an operator can fix it.
    `switched_off` — the tenant turned it off (sentiment and emotions only).
    `no_target_language` — translation is configured but neither the tenant's
    `target_language` nor the deployment default resolves (translation only; it has no
    on/off switch, so an absent target is its off state).

    - `"not_configured"`

    - `"switched_off"`

    - `"no_target_language"`

- `tenant_id: string`

- `translation: TypeStatus`

  One enrichment's progress for a tenant. When `enabled` is false, `eligible` and `done` are zero.

  - `done: number`

    Eligible records that have been enriched.

  - `eligible: number`

    Feedback records that qualify for this enrichment.

  - `enabled: boolean`

    Whether the enrichment is active for the tenant (deployment-configured and switched on / with a resolvable target language).

  - `failed: number`

    Eligible records whose last enrichment attempt gave up but which a retry could still
    rescue — a provider outage, a timeout. Counted only while the record is still
    un-enriched, so a later success removes it without any cleanup.

  - `failed_terminal: number`

    Eligible records the provider will never accept, because the outcome is a property of
    the record's own text — a content-policy block, a refusal, an input past the model's
    limit. Retrying these cannot help; they resolve only if the text changes.

  - `disabled_reason: optional "not_configured" or "switched_off" or "no_target_language"`

    Which gate switched the enrichment off. Present exactly when `enabled` is false, and
    absent otherwise. `not_configured` — the deployment has no provider/model for this
    enrichment, so it is off for every tenant and only an operator can fix it.
    `switched_off` — the tenant turned it off (sentiment and emotions only).
    `no_target_language` — translation is configured but neither the tenant's
    `target_language` nor the deployment default resolves (translation only; it has no
    on/off switch, so an absent target is its off state).

    - `"not_configured"`

    - `"switched_off"`

    - `"no_target_language"`

### Example

```http
curl http://localhost:8080/v1/enrichment-status \
    -H "Authorization: Bearer $HUB_API_KEY"
```