> ## Documentation Index > Fetch the complete documentation index at: https://docs.anthid.com/llms.txt > Use this file to discover all available pages before exploring further. # Report which simulated accounts are actually running > Performs, once and on demand, the comparison the scheduler makes every second: what should be running, against what the connection bucket says is. `observed` counts every tenant's runners by state and carries no tenant. `test_accounts` lists the caller's simulated accounts, each with what the bucket says about it, and only simulated accounts: for a live account the scheduler additionally requires the organization's live trading entitlement, which this service does not check, so comparing one here would report a billing exclusion as a missing runner. Bucket entries expire sixty seconds after the last heartbeat, so an absent runner means nothing has reported within the minute rather than that none exists, and `checked_at` is when the bucket was read. Readable by a client credential holding the credential:read scope, which sees every organization's simulated accounts, and by an organization admin of a configured internal organization, which sees its own. ## OpenAPI ````yaml /api-reference/accounts.openapi.json get /v1/accounts/runners/health openapi: 3.1.0 info: title: accounts description: '' license: name: '' version: 0.1.0 servers: - url: https://api.anthid.com description: Production security: [] tags: - name: Accounts description: Create, inspect, update, and disable trading accounts - name: Broker Credentials description: Inspect and manage credentials linked to trading accounts paths: /v1/accounts/runners/health: get: tags: - Accounts summary: Report which simulated accounts are actually running description: >- Performs, once and on demand, the comparison the scheduler makes every second: what should be running, against what the connection bucket says is. `observed` counts every tenant's runners by state and carries no tenant. `test_accounts` lists the caller's simulated accounts, each with what the bucket says about it, and only simulated accounts: for a live account the scheduler additionally requires the organization's live trading entitlement, which this service does not check, so comparing one here would report a billing exclusion as a missing runner. Bucket entries expire sixty seconds after the last heartbeat, so an absent runner means nothing has reported within the minute rather than that none exists, and `checked_at` is when the bucket was read. Readable by a client credential holding the credential:read scope, which sees every organization's simulated accounts, and by an organization admin of a configured internal organization, which sees its own. operationId: getRunnerHealth responses: '200': description: What is running, against what should be content: application/json: schema: $ref: '#/components/schemas/RunnerHealthResponse' example: checked_at: '2026-08-26T12:00:00Z' missing: 1 observed: active: 12 deleting: 0 pending: 1 stranded: 0 test_accounts: - enabled: true name: load-test-01 state: active trading_account_id: 550e8400-e29b-41d4-a716-446655440111 - enabled: true name: load-test-02 state: absent trading_account_id: 550e8400-e29b-41d4-a716-446655440222 '401': description: Caller is an API key, or a user whose token carries no organization content: text/plain: schema: type: string '403': description: >- The client credential is missing the credential:read scope, the caller's organization is not configured as internal, or the caller is not an organization admin content: text/plain: schema: type: string '429': description: Caller exceeded the request rate limit content: text/plain: schema: type: string '500': description: Unexpected server error while reading the connection bucket content: text/plain: schema: type: string security: - bearerAuth: [] components: schemas: RunnerHealthResponse: type: object description: The reconciliation the scheduler performs every second, reported once. required: - observed - test_accounts - missing - stranded - checked_at properties: checked_at: type: string format: date-time description: >- When the bucket was read. Returned rather than left to the caller's clock because every state here is relative to the bucket's sixty second expiry, and a reader working out staleness from its own `now` would be measuring its own latency into the answer. missing: type: integer format: int64 description: Enabled accounts in `test_accounts` that nothing is running. observed: $ref: '#/components/schemas/ObservedRunners' description: Every tenant's runners, counted by state. No tenant in it. stranded: type: integer format: int64 description: Disabled accounts in `test_accounts` that something still is. test_accounts: type: array items: $ref: '#/components/schemas/TestAccountRunner' description: >- The caller's simulated accounts, each with what the bucket says. Test accounts only, and that is a correctness boundary rather than an omission. See the handler for why real trading accounts are not compared here. ObservedRunners: type: object description: >- How many accounts the bucket is holding in each state, across every tenant. Counts and nothing else. No account id, no organization: this is the "is the fleet up at all" figure, and it carries no tenant for the same reason the outbox health does. See [`can_read_test_accounts`] for the line. [`can_read_test_accounts`]: crate::policy::can_read_test_accounts::can_read_test_accounts required: - active - pending - deleting properties: active: type: integer format: int64 deleting: type: integer format: int64 pending: type: integer format: int64 TestAccountRunner: type: object description: One simulated account, and whether anything is running it. required: - trading_account_id - name - enabled - state properties: enabled: type: boolean description: |- Whether the account is registered to run at all. A disabled account being `Absent` is the correct outcome, not a fault. name: type: string state: $ref: '#/components/schemas/RunnerState' trading_account_id: type: string format: uuid RunnerState: type: string description: >- What the connection bucket says about one account, right now. Observed rather than authoritative, and the distinction is load bearing. The bucket's entries expire after sixty seconds unless a runner's heartbeat refreshes them, so [`RunnerState::Absent`] means "nothing has heartbeated for this account within the last minute", not "no runner exists". A reader must not treat this as the desired state's contradiction; it is the last thing anybody saw. enum: - active - pending - deleting - absent securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: JWT ````