---
title: "Get Instance Health"
method: GET
path: "/api/integrations/instances/{instance_id}/health"
tags: ["integrations"]
---

# Get Instance Health

`GET /api/integrations/instances/{instance_id}/health`

## Request

- Base URL: `https://app.thinqos.com` (the document declares no server; this is the origin it was published from)
- URL: `https://app.thinqos.com/api/integrations/instances/{instance_id}/health`
- Auth: none declared

Health detail for one connection (TOS-2708).

Reads the persisted rollup — it does NOT probe. The detail view is opened
far more often than health actually changes, and probing on read would turn
every page load into an upstream call.

## Path parameters

- `instance_id` string, uuid, required

## Headers

- `x-impersonate-identity-id` string, nullable

## Response `200`

Successful Response

- ConnectionHealthOut: Health of one connection, rendered without the client inventing anything. ``status`` carries the Connected-filter buckets directly (``ready`` / ``degraded`` / ``needs_attention`` / ``unknown``), so the list filters and the degraded/failure screens all read one field. ``credential_status`` keeps the underlying evidence-ladder value for callers that need it. ``unknown`` means CONNECTED BUT NOT YET MEASURED — a credential-free integration that owns no connection row, or a connection probed before the rollup existed. There is no evidence of a problem, so it belongs in the **Ready** filter bucket; render the detail as "Not checked yet" rather than showing ``summary`` (which would read "0 of 0 tools available"). It is a distinct value rather than being folded into ``ready`` because claiming "no capability is failing" when nothing was measured is a fabrication.
  - `status` 'ready' | 'degraded' | 'needs_attention' | 'unknown', required
  - `credential_status` 'connected' | 'needs_reconnect' | 'admin_repair' | 'unreachable' | 'missing', required
  - `healthy_tools` integer, required
  - `total_tools` integer, required
  - `affected_tools` AffectedToolOut[]
    - `name` string, required
    - `reason` string, required
    - `failure_code` string, nullable
    - `evidence` 'probe' | 'telemetry', required
    - `observed_at` string, date-time, nullable
  - `healthy_summary` string, required
  - `summary` string, required
  - `checked_at` string, date-time, nullable
  - `failure` ConnectionFailureDetail: The four named fields the connection-failure screen renders, plus the action. Populated from the closed ``IntegrationFailureCode`` taxonomy (TOS-2708), so every failure category — OAuth denial, wrong account, invalid/expired key, missing scopes, failed safe probe, rate limit, unreachable endpoint, partially unavailable tools — resolves to named copy rather than an error string the user has to interpret.
    - `code` string, required
    - `failed_check` string, required
    - `cause` string, required
    - `preserved` string, required
    - `diagnostic_reference` string, required
    - `next_action` string, required

## Other responses

- `422`: Validation Error

---

[API](https://skmtc.dev/thinqos/apis/thinqos-api.md) · [All operations](https://skmtc.dev/thinqos/apis/thinqos-api/llms.txt) · [OpenAPI document](https://skmtc.dev/thinqos/apis/thinqos-api/revisions/ce2c2ce318b1?raw)