> ## Documentation Index > Fetch the complete documentation index at: https://docs.arlohealth.ai/llms.txt > Use this file to discover all available pages before exploring further. # Wait for Reply > Blocks until the next reply lands in the conversation (an AI triage reply, a provider connecting, or a provider message) and returns it, instead of polling GET /api/consultations/{id}. The REST form of the wait_for_reply tool. Resumable: when nothing arrives inside the window the response is `stillWaiting: true`; call again with the same parameters. Pass the last reply's messageId as `afterMessageId` to await what comes after it. In IDLE or PAYMENT_REQUIRED there is nothing to wait for and the response says so with a `nextStep`. ## OpenAPI ````yaml api-reference/openapi.yaml GET /api/consultations/{id}/reply openapi: 3.1.0 info: title: Arlo Health MCP Server version: 1.0.0 description: | Healthcare infrastructure for AI agents. Everything an agent needs to act in healthcare for its user. ## Overview This server exposes two interfaces: - **MCP** (Model Context Protocol) over streamable HTTP at `/` for AI agents - **REST API** at `/api/*` for bots, skills, and direct HTTP clients Both interfaces use the same tool handlers and authentication. ## Authentication OAuth 2.1 with PKCE. All `/api/*` endpoints require a Bearer token. Discovery endpoints (no auth required): - `/.well-known/oauth-protected-resource` - `/.well-known/oauth-authorization-server` - `/.well-known/mcp.json` (MCP Server Card) ## Service Regions Care is licensed by where the patient is physically located when they request it, so every conversation carries an ISO 3166-2 `region` code. Arlo currently serves California, United States (`US-CA`); an unsupported region is refused with `REGION_NOT_SUPPORTED`, whose payload names the regions Arlo serves at that moment. contact: name: Arlo Health url: https://arlohealth.ai license: name: Proprietary url: https://arlohealth.ai/tos servers: - url: https://mcp.arlohealth.ai description: Production MCP Server security: - oauth2: - openid - profile - email - offline_access tags: - name: Discovery description: MCP and OAuth discovery endpoints (no auth required) - name: Profile description: User profile and patient information - name: Consultations description: Healthcare consultation management - name: Messaging description: Conversation messaging and media - name: Prescriptions description: Prescription orders (read-only; Photon Health fulfills them) - name: Health Records description: The user's connected insurance and health records (US) - name: Care Pricing description: Free price transparency under the user's plan (US) - name: Care Jobs description: >- Real-world care actions Arlo's care team executes after the patient approves (US) - name: Payment description: Billing and pay-per-use payment management - name: Webhooks description: Webhook registration and status paths: /api/consultations/{id}/reply: get: tags: - Messaging summary: Wait for the next reply description: | Blocks until the next reply lands in the conversation (an AI triage reply, a provider connecting, or a provider message) and returns it, instead of polling GET /api/consultations/{id}. The REST form of the wait_for_reply tool. Resumable: when nothing arrives inside the window the response is `stillWaiting: true`; call again with the same parameters. Pass the last reply's messageId as `afterMessageId` to await what comes after it. In IDLE or PAYMENT_REQUIRED there is nothing to wait for and the response says so with a `nextStep`. operationId: waitForReply parameters: - name: id in: path required: true schema: type: string description: Consultation/conversation ID - name: afterMessageId in: query required: false schema: type: string description: Await the reply after this message (the last one you have seen) - name: maxWaitSeconds in: query required: false schema: type: integer minimum: 5 maximum: 300 default: 55 description: >- How long one call may block. Defaults to 55 so a plain HTTP client's 60-second timeout is never hit; clamped to 300. responses: '200': description: The next reply, several pending replies, or stillWaiting content: application/json: schema: $ref: '#/components/schemas/WaitReply' '400': $ref: '#/components/responses/BadRequest' '401': $ref: '#/components/responses/Unauthorized' components: schemas: WaitReply: type: object description: | One of: a single reply, several pending replies, `stillWaiting` (resume with the same call), `connected` (a provider joined), or a note that the conversation has nothing pending to wait for. properties: conversationId: type: string status: 192865e6-ea37-4719-937e-8ec96bfd8583 reply: $ref: '#/components/schemas/WaitEvent' replies: type: array description: >- Several replies had already arrived. Relay them in order, then call again with afterMessageId set to the LAST reply's messageId. items: $ref: '#/components/schemas/WaitEvent' stillWaiting: type: boolean description: >- True when nothing arrived within the window. Call again with the same parameters to resume; no duplicate side effects. waitedSeconds: type: integer description: With stillWaiting, how long this call blocked. connected: type: boolean description: >- True when a provider joined while waiting. Fetch the conversation, then message the provider. note: type: string description: >- Why there was nothing to wait for (IDLE, PAYMENT_REQUIRED, EMERGENCY). nextStep: type: string nextStepArgs: type: object WaitEvent: type: object description: A single inbound message or visit event. properties: from: type: string enum: - ai - provider type: type: string description: text, photo, file, prescription, lab_request, or a participant join text: type: string messageId: type: string description: >- Pass as afterMessageId to await the next reply, and as lastSeenMessageId when sending. fileName: type: string prescriptionOrderId: type: string timestamp: type: string Error: type: object description: Error response properties: error: type: string description: Error code reason: type: string description: Human-readable explanation code: type: string description: Detailed error code responses: BadRequest: description: Invalid request content: application/json: schema: $ref: '#/components/schemas/Error' Unauthorized: description: Missing or invalid authentication content: application/json: schema: type: object properties: error: type: string needsAuth: type: boolean securitySchemes: oauth2: type: oauth2 description: OAuth 2.1 with PKCE flows: authorizationCode: authorizationUrl: /oauth/authorize tokenUrl: /oauth/token scopes: openid: OpenID Connect profile: User profile email: Email address offline_access: Refresh tokens ````