Zendesk

Zendesk's Support API is the most-integrated helpdesk API in production — agents building customer-facing workflows (ticket routing, auto-replies, escalation logic, analytics) will encounter it frequently. The API covers tickets, users, organizations, views, macros, triggers, automations, and the knowledge base. Sunshine platform extends the data model with custom objects and relationships. For agents, the ticket lifecycle (create → update → solve → close) is well-modeled with webhooks and triggers for event-driven automation. The main challenges: rate limits scale with plan tier and can throttle agents on lower plans, the API surface is large with inconsistent naming conventions across older and newer endpoints, and the Zendesk marketplace OAuth flow adds complexity for multi-tenant agent apps.

Three auth methods: API token (email/token:{api_token} as Basic Auth — simplest for agents), OAuth (required for Zendesk Marketplace apps), and password auth (deprecated but still functional). API tokens are generated per-admin and grant full API access — no permission scoping. OAuth supports authorization code flow with scoped access (read, write, impersonate). For agents embedded in customer support workflows, the API token is the quickest path. Multi-tenant agents serving multiple Zendesk instances should use OAuth for proper credential isolation. IP restriction is available on higher plans. SAML/SSO for console access. No built-in key rotation mechanism — admins must manually create new tokens and revoke old ones.

REST API at {subdomain}.zendesk.com/api/v2/ with conventional CRUD on resources. Tickets are the central object: create, update, bulk update, merge, and close. Side-loading (include=users,organizations) reduces API calls by embedding related resources. Incremental exports enable agents to sync ticket data efficiently — cursor-based pagination over changes since a timestamp. The Search API supports Zendesk Query Language for complex ticket queries across any field. Webhook-based event notifications via targets and webhooks. The API has grown organically over 10+ years, leading to some inconsistencies: some endpoints use JSON envelope patterns, others return arrays directly. Pagination varies between cursor-based (newer) and offset-based (older endpoints).

API errors return JSON with error and description fields, sometimes with details object for field-level validation errors. The most impactful operational constraint is rate limiting: 400 requests/minute on Team plan, 700 on Professional, and custom limits on Enterprise. Exceeding limits returns 429 with Retry-After header. Agents doing bulk operations (ticket migration, data sync) must implement throttling. Job statuses for bulk operations (bulk ticket update) are async — agents poll a job status endpoint. The incremental export API has a 1-request-per-second limit. Webhook delivery retries failed deliveries up to 3 times with exponential backoff. Deprecation notices for API changes are communicated through the developer blog and changelog.

Developer documentation at developer.zendesk.com is comprehensive, covering REST API, SDKs, Sunshine platform, and marketplace development. API reference includes endpoint descriptions, parameter tables, and response examples. The documentation distinguishes between Support, Chat, Talk, and Guide APIs clearly. ZAF (Zendesk Apps Framework) documentation covers building embedded apps. SDKs are available for Ruby, Python, Node, PHP, Java, and Go — varying maintenance levels. The community forums and Stack Overflow coverage are strong due to Zendesk's market penetration. The main documentation weakness: finding the right guide for a specific integration pattern requires navigating a large doc surface, and some guides still reference older API patterns alongside current ones.