Short answer
The Jobber API is a single GraphQL endpoint covering clients, requests, quotes, jobs, visits and invoices, with OAuth 2.0 authorization code and 60-minute tokens. Apps are registered in Jobber's Developer Center; Marketplace apps go through review, and unreviewed Draft apps are blocked beyond 5 paying Jobber accounts. Limits are 2,500 requests per 5 minutes plus a GraphQL query-cost budget. If the goal is an AI assistant working with field service data rather than a system-to-system integration, an MCP server is the faster route.
Authentication
OAuth 2.0 authorization code. Register an app in the Jobber Developer Center (name, scopes, callback URL) to get a client ID and secret; a Jobber admin authorizes it. Access tokens expire after 60 minutes by default and refresh.
Getting access
- A free Developer Center account plus a Jobber developer testing account (90 days, extendable).
- Publishing to the App Marketplace requires Jobber's app review, typically including a beta with selected customers.
- Custom integrations can stay in Draft without review, but API access is blocked if a Draft app connects to more than 5 paying Jobber accounts unless Jobber approves.
- Which Jobber plans can connect apps, and any marketplace fees: not published in the developer docs.
What you can reach
- Clients and properties
- Requests and quotes
- Jobs and visits (scheduling, assigned users)
- Invoices
- Users (team members)
- Products and services
- Expenses and time sheet entries
- Assessments and custom fields
Webhooks, rate limits and test environment
| Jobber | |
|---|---|
| Webhooks | Yes: signed POSTs (HMAC-SHA256 with the app's secret), at-least-once delivery, examples include CLIENT_CREATE and CLIENT_UPDATE. Payloads carry only IDs, so you query the API for details, and your endpoint must answer within 1 second. |
| Rate limits | 2,500 requests per 5 minutes per app and account, plus a GraphQL query-cost limit of 10,000 points restoring at 500 points per second. |
| Sandbox / test environment | No separate sandbox; developers use a developer-testing Jobber account and the GraphiQL explorer. |
Known limitations
From Jobber's published documentation:
- Draft apps are blocked beyond 5 paying Jobber accounts
- Adding scopes to a published app forces every connected account to re-authorize
- Webhook payloads are IDs only and must be acknowledged within 1 second
- The schema is only browsable in GraphiQL after you create an app
Building on the API versus connecting AI through MCP
If you want an integration (syncing Jobber with another system), the API is the route. If you want your team to ask an AI assistant about jobs, schedules and invoices, or have it do the work, you would be building and hosting that yourself on top of the API.
Fieldproxy takes the other route: its official MCP server gives Claude, ChatGPT and Microsoft Copilot 145 field service tools directly, with scoped access and a confirm step before every change, and it can also change the software itself (fields, screens, automations) in a sandbox.
| Build on the Jobber API | Fieldproxy MCP server | |
|---|---|---|
| Time to first answer in ChatGPT or Claude | Build, host and maintain a server | Add a connector and sign in |
| Auth and permissions | You implement them | Per-person sign-in, scopes, row and column limits |
| Changes to data | Raw API writes | Proposed, confirmed by a person, undoable |
| Changing the software itself | Not available through an API | Fields, screens, automations, in a sandbox |
Sources
- Jobber Developer Center
- App authorization (OAuth 2.0)
- API rate limits
- Setting up webhooks
- Custom integrations
Read from the vendor's own developer documentation on September 27, 2026. Anything the docs do not publish is marked "not published".
Fieldproxy facts on this page come from the Fieldproxy MCP reference. Competitor details are from each vendor's own documentation as of September 27, 2026; see the comparison table and sources.