> ## Documentation Index > Fetch the complete documentation index at: https://docs.fonoster.com/llms.txt > Use this file to discover all available pages before exploring further. # Node SDK > Every client class and method in @fonoster/sdk: what it does, the parameters it takes and what it returns. `@fonoster/sdk` manages a workspace from Node.js or the browser: applications, numbers, trunks, SIP agents, calls, call records, webhooks, providers and keys. It wraps Fonoster's [gRPC API](https://docs.fonoster.com/reference/api.md). To answer calls with your own code, use the [Voice SDK](https://docs.fonoster.com/reference/voice-sdk.md) instead. ## Install ```bash theme={"system"} npm install @fonoster/sdk ``` The package works in Node.js and in browsers. In the browser it talks grpc-web to the same endpoint. ## Create a client and sign in Create a client with your workspace's access key ID, then sign in with an API key: ```javascript theme={"system"} const SDK = require("@fonoster/sdk"); const client = new SDK.Client({ accessKeyId: "WO7d2kq9x4m1c8b3n6p0r5t2v9y4z7a1e3" }); await client.loginWithApiKey( process.env.FONOSTER_API_KEY, // APk3n8q2w7e5r1t9y4u6i0o3p8a2s5d7f1 process.env.FONOSTER_API_SECRET ); const apps = new SDK.Applications(client); ``` The client's `accessKeyId` selects the workspace: every request goes to the workspace whose access key ID starts with `WO`. The API key ID (`AP…`) and its secret prove who you are. See [API keys and access](https://docs.fonoster.com/workspace/access.md) for where to find each value. ### `Client` options (Node.js) **accessKeyId** `string` *required* The workspace's access key ID (`WO` and 32 characters). Every request acts on this workspace. **endpoint** `string` (default: api.fonoster.ai) The gRPC endpoint, as `host` or `host:port`. Set it to your install's address when you self-host. **allowInsecure** `boolean` (default: false) Connect without TLS. Use it only for a local install on plain HTTP. **httpUrl** `string` The base URL of the HTTP routes (call records and regions). Default `https://` plus the endpoint's host, without the port. **withoutInterceptors** `boolean` (default: false) Turn off automatic token renewal. ### `WebClient` options (browser) The browser build exports `WebClient` instead of `Client`. It takes `url` (default `https://api.fonoster.ai`) and `accessKeyId`, and has the same sign-in methods. ### Sign-in and session methods These are methods of `Client` and `WebClient`. Each sign-in method stores the tokens on the client; the client renews the access token before a request when it has expired. | Method | Purpose | | - | - | | `loginWithApiKey(accessKeyId, accessKeySecret)` | Sign in with an API key. The usual choice for code | | `login(username, password, twoFactorCode?)` | Sign in as a user with email and password | | `loginWithRefreshToken(refreshToken)` | Sign in with a refresh token you kept | | `loginWithVerificationCode(email, code)` | Sign in with the 6-digit code from `sendVerificationCode("EMAIL", email)` | | `loginWithOauth2Code("GITHUB", code)` | Sign in with a GitHub OAuth code | | `logout()` | Forget the tokens and revoke the refresh token | | `refreshToken()` | Renew the access token now | | `setAccessKeyId(accessKeyId)` | Switch to another workspace you belong to | | `getAccessKeyId()`, `getAccessToken()`, `getRefreshToken()`, `getIdToken()` | Read the current values | | `getCapabilities()` | What the deployment supports: sign-in methods, sign-up policy, billing, regions and SIP connection details. Needs no sign-in | ## Conventions * Every method returns a promise. A rejected gRPC call rejects with the gRPC error. * **Refs** are UUIDs, such as `3f8b2c71-9d4e-4a6b-8c25-7e1f0a9b4d62`. `create*`, `update*` and `delete*` methods resolve to `{ ref }` unless noted. * **List methods** take `{ pageSize, pageToken }` and resolve to `{ items, nextPageToken }`. Pass `nextPageToken` back as `pageToken` to get the next page. * **Dates** are `Date` objects unless noted. * `CallRecords` and `Regions` use HTTP. They reject with `CallRecordsError` or `RegionsError`, which carry `status` (the HTTP status) and `code` (the error code from the body). ## Applications An application is a stream (`STREAM`) or an Autopilot agent (`AUTOPILOT`). `new SDK.Applications(client)`. | Method | Purpose | Returns | | - | - | - | | `createApplication(request)` | Create a stream or an agent | `{ ref, signingSecret, signingSecretHint }`. The secret is shown only here and on rotation; Autopilot agents get none | | `getApplication(ref)` | Get one application | `Application` | | `updateApplication({ ref, ...fields })` | Replace the application; fields you leave out reset to their defaults. Read it with `getApplication`, change what you need, and send all of it back | `{ ref }`, plus `signingSecret` when the update generated a new one | | `listApplications({ pageSize, pageToken })` | List applications | `{ items: Application[], nextPageToken }` | | `deleteApplication(ref)` | Delete an application | `{ ref }` | | `rotateSigningSecret(ref)` | Replace a stream's signing secret | `{ ref, signingSecret, signingSecretHint }` | | `evaluateIntelligence({ intelligence: { config } })` | Run an agent's test cases | An async iterator of events: `stepResult`, `scenarioSummary`, `evalError` | | `createTestToken()` | Get a short-lived token for a test call from the browser | `{ domain, displayName, signalingServer, targetAor, username, token }` | The request for `createApplication`: **name** `string` *required* A name for the application. **type** `"STREAM" | "AUTOPILOT"` *required* A stream runs your WebSocket server. An Autopilot agent runs on Fonoster. **url** `string` Streams: the `ws://` or `wss://` address Fonoster dials for each call. **protocol** `"fonoster" | "twilio"` (default: fonoster) The stream's wire protocol: the [session protocol](https://docs.fonoster.com/reference/session-protocol.md) or [Twilio Media Streams](https://docs.fonoster.com/reference/twilio-compatibility.md). **speech** `object` Speech defaults. Voices and models are catalog ids, such as `deepgram/aura-2-thalia-en`. **properties** **voice** `string` Default voice. **language** `string` Default language, such as `en-US`. **sttModel** `string` Speech-to-text model. **ttsModel** `string` Text-to-speech model. **silenceTimeoutMs** `number` Silence timeout in milliseconds. **patience** `"low" | "normal" | "high"` How long the turn detector waits before it ends the caller's turn. **audio** `object` Audio format for `fonoster` streams: `{ codec: "pcm16" | "pcmu", rate }`. **fallbackAppRef** `string` Another application to take the call when this one can't be reached. **fallbackMessage** `string` Text spoken before hanging up when the application can't be reached. **listeners** `object[]` Listen-only WebSockets: `{ url, tracks: "inbound" | "outbound" | "both" }`. Up to 4 are dialed. **recording** `"off" | "on" | "consent" | ""` Recording policy. Empty inherits the workspace's policy. See [Recording](https://docs.fonoster.com/telephony/recording.md). **amd** `object` Answering-machine detection: `{ enabled, windowMs }`. `windowMs` defaults to 4000 (500 to 15000\). Off when unset. See [Answering-machine detection](https://docs.fonoster.com/telephony/amd.md). **intelligence** `object` Autopilot agents: `{ config }`, the agent's configuration. See [Autopilot configuration](https://docs.fonoster.com/reference/autopilot-config.md). An `Application` has the same fields, plus `ref`, `signingSecretHint` (the secret's last characters), `createdAt` and `updatedAt`. ## Numbers Phone numbers and where their calls go. `new SDK.Numbers(client)`. | Method | Purpose | Key params | Returns | | - | - | - | - | | `createNumber(request)` | Add a number | `name`, `telUrl` (`tel:+18095550123`), `city`, `country`, `countryIsoCode`, `trunkRef`, and one destination: `appRef` or `agentAor` | `{ ref }` | | `getNumber(ref)` | Get one number | | `{ ref, name, telUrl, appRef, agentAor, city, country, countryIsoCode, trunk: { ref, name }, createdAt, updatedAt }` | | `updateNumber({ ref, ...fields })` | Change the name, trunk or destination | `name`, `trunkRef`, `appRef`, `agentAor` | `{ ref }` | | `listNumbers({ pageSize, pageToken })` | List numbers | | `{ items, nextPageToken }` | | `deleteNumber(ref)` | Remove a number | | `{ ref }` | `telUrl`, `city`, `country` and `countryIsoCode` can't change after creation. ## Trunks SIP connections to your carriers. `new SDK.Trunks(client)`. | Method | Purpose | Key params | Returns | | - | - | - | - | | `createTrunk(request)` | Add a trunk | `name`, `inboundUri`, `sendRegister`, `accessControlListRef`, `inboundCredentialsRef`, `outboundCredentialsRef`, `uris` | `{ ref }` | | `getTrunk(ref)` | Get one trunk | | `Trunk` | | `updateTrunk({ ref, ...fields })` | Change any field | same as create | `{ ref }` | | `listTrunks({ pageSize, pageToken })` | List trunks | | `{ items, nextPageToken }` | | `deleteTrunk(ref)` | Remove a trunk | | `{ ref }` | Each entry in `uris` is the carrier's address for outbound calls: `{ host, port, transport, user, weight, priority, enabled }`. `transport` is one of `UDP`, `TCP`, `TLS`, `SCTP`, `WS`, `WSS`. ## Domains SIP domains that SIP agents register to. `new SDK.Domains(client)`. | Method | Purpose | Key params | Returns | | - | - | - | - | | `createDomain(request)` | Add a domain | `name`, `domainUri`, `accessControlListRef`, `egressPolicies` (`[{ rule, numberRef }]`) | `{ ref }` | | `getDomain(ref)` | Get one domain, with its ACL | | `Domain` | | `updateDomain({ ref, ...fields })` | Change the name, ACL or egress policies | | `{ ref }` | | `listDomains({ pageSize, pageToken })` | List domains | | `{ items, nextPageToken }` | | `deleteDomain(ref)` | Remove a domain | | `{ ref }` | `domainUri` can't change after creation. Names are one label under your install's roots: `