LiveChat API Guide: Text API Tokens, Quotas, Webhooks and MCP

LiveChat, now sold to new customers as Text plans, gives developers one Text API for chats, tickets, configuration, reports and catalogs. Every paid Text plan includes API calls: 10,000 a month on Essential and 20,000 a month on Growth, reset each billing cycle, with an extra pack of 100,000 calls at $29.50 ^[1]. Authentication is OAuth 2.1, and a personal access token gets you to a first request fastest. Our verdict: the platform is mature and well versioned. The quota is the thing to plan. Count your calls before you build, because almost every request counts.

What the Text API covers

Text splits its developer platform into three families. The table below maps each one to the jobs it handles, so you can find the right reference before writing code. The guides send requests to api.livechatinc.com with the version in the path, and the same Account and organization model sits under every family: one user has one Text account that can belong to several organizations, and most endpoints identify the organization by its ID.

Each messaging API has a request-and-response Web API, and the two chat APIs also offer an RTM WebSocket variant for persistent connections with pushed events. Bots use the Web API only and behave like human agents at the protocol level. An API bot is your own automated participant; it is not the AI Agent you configure in the Text app.

Versions follow a fixed lifecycle. Agent Chat, Customer Chat, Configuration, Reports and Webhooks sit at v3.6 stable (since 7 August 2025), v3.7 developer preview (since 3 July 2026) and v3.5 legacy. Text says to build new integrations on the stable version. A legacy version still answers but gets only critical and security fixes, and its responses carry a Legacy header with the end date. A deprecated version carries a Deprecation header and stops working when that date passes. The Customer Data Platform, Partner and Global Accounts APIs run one continuously updated version instead. The Chat Widget JS API is on v5.0 stable, with v4.0 legacy.

Tokens, scopes and the OAuth flows

Start with a personal access token. You create it in the Text app under Settings, API access, pick its scopes, and send it as Basic auth with your Account ID as the username ^[2]. It always acts with the permissions of the person who created it. That suits an internal script. It does not suit a tool that acts for several users.

For apps that other people install, use OAuth 2.1. Text supports three flows. The authorization code grant is the recommended one: Web app clients must use PKCE, while server-side apps use a client secret. Sign in with Text fits internal tools for one organization, and the implicit grant still works but issues no refresh token. OAuth access tokens expire after 8 hours (28,800 seconds), and a refresh token renews them without a new login. Each client is limited to 20 active tokens per user; the oldest is revoked when you go over.

Two rules catch newcomers. Agent tokens and customer tokens are separate: an agent token fails on the Customer Chat API, and a customer token works nowhere else. Customer tokens come from their own grants (cookie, agent token or identity token) and also last 28,800 seconds by default. Scopes are fixed when a token is created and cannot be widened later, so plan them first. Each scope is read only (ro), read and write (rw) or read and create (rc), with reach modifiers my, groups, access and all. What a user may request also depends on two separate roles, the Global Accounts role and the Text role.

How the API quota is counted on Text plans

The subscription bills seats per user per month, while API usage is measured separately in calls a month ^[3]. The rule for counting is strict. Every request to Text APIs and SDKs counts, including calls made with personal access tokens, and each call counts the same whichever API it hits. Failed requests count too, except those that return HTTP 500. You can track request volume in the Billing section of the Text app settings.

One setup multiplies usage quietly. A Custom Identity Provider routes all widget traffic through the Customer Chat API, so every one of those requests counts against the quota. Polling is the other trap. A service that checks for new chats once a minute sends 1,440 requests a day, or 43,200 in a 30-day month by our arithmetic. That passes the Growth allowance on its own. Webhooks push events to your server instead, so you fetch data only when something happens.

The arithmetic for going over is simple. A Growth account that needs 30,000 calls a month buys one pack: $29.50 covers up to 100,000 extra calls. Upgrading for calls alone makes no sense. Growth costs $60 more per user per month billed annually than Essential yet adds only 10,000 calls a month, while one pack adds ten times as many for less. Plan choice should follow features like routing, not the API. Seats are a separate cost covered in our LiveChat pricing guide.

What platform.text.com lists for private API usage

Text's developer platform keeps its own pricing page for private API usage, meaning calls from private apps and personal access tokens ^[4]. That page lists tiered rates of $0.27 per 1,000 requests up to 1 million requests a month, $0.22 per 1,000 from 1 million to 10 million, and $0.17 per 1,000 above 10 million. It also lists two prepaid API packages, $29.99 a month for 100,000 requests and $149.99 a month for 500,000, which are platform packages, not Text plans. Pro+ costs the same as Pro and adds a one-time, one-hour session with an API specialist.

The same page sets rules for those packages. Only one package can be active at a time, so buying two gives you only the larger one's requests. Packages renew each month until you uninstall them, unused requests reset at the end of the period, and going over a package does not block requests. Calls from apps installed from the public Marketplace do not count toward private usage; you pay only the app's own subscription. AI resolutions are a separate Text meter and do not draw on API calls.

Webhooks, widget SDKs and catalog APIs

Webhooks are registered through the Configuration API and send an HTTP request to your URL when an event fires, such as incoming_chat, chat_deactivated or ticket_created. They arrive from the IP range 185.177.172.0/22, so allow that range on your firewall. A webhook can carry the full change or just enough data to fetch the affected resource. An incoming_chat webhook includes the events that started the first thread.

For the customer side, Text offers three widget tools. The Customer SDK lets you build a chat widget from scratch, and Text says its own native widget runs on the same SDK with no private APIs. The Chat Widget JS API controls the standard widget, for example setting session variables or a customer's cart. Adapters wrap it for React, Vue and Angular. Our website widget guide covers the no-code install paths.

Two newer APIs feed commerce features. The Product Directory API, added on 19 August 2026, uploads a JSON catalog of up to 50 MB, one catalog per organization, so the AI Agent can show product cards. The Sales Tracking API reports orders from any store stack through one call on the order confirmation page. The Reports API exposes the dashboard's performance data, and its totals are aggregates, so per-agent values can add up to more than the overall total.

The MCP server and Text's reference guides

Text hosts an MCP server at mcp.text.com on every plan. It connects AI assistants to your workspace with OAuth, the recommended option, or a bearer token. The server gives access to tickets and chat transcripts under the connected user's permissions, so an assistant sees exactly what that user can see. Text publishes setup guides for Claude Desktop, Claude Code, Claude Web and ChatGPT, and no code is needed on your side.

Text also publishes sample projects: Salesforce chat routing, which needs a Text account on Growth or higher, HubSpot contact sync, Shopify orders to tickets, a Metabase dashboard and a Google Sheets reports template. Text marks the routing and sync samples as not production ready. The Shopify sample keeps deduplication in memory, so a restart can create duplicate tickets. The HubSpot sample can lose a sync on restart. Treat them as starting code, not finished integrations. For ready-made connections that need no code, see our LiveChat integrations guide.

The bottom line

The Text API is a full, versioned platform: three API families, clear token rules, webhooks, widget SDKs and an MCP server on every plan. The cost question is narrow. Essential includes 10,000 calls a month and Growth 20,000, and a pack of 100,000 extra calls costs $29.50. Design around webhooks rather than polling, give every token only the scopes it needs, and build on v3.6 stable. Then the quota rarely becomes the constraint.