Is there an MCP server for Zendesk?

Yes, several. Swifteq publishes a supported one on the Zendesk Marketplace, and there are community servers on GitHub. All of them wrap the ticket API: fetch a ticket, search tickets, read comments. Zendesk's own MCP support is the client side, letting Zendesk call your tools rather than letting your AI client read its data. Rulebase is a different thing: it syncs the conversations into one normalised record you can sample, join across channels and measure, with 149 customer-operations skills on top.

Do I have to give an AI agent write access to Zendesk?

No. All but one skill in the library is read-only, and the MCP server reads from Rulebase rather than acting on your helpdesk. Anything that would change a record is marked as such and asks before it writes.

What can an AI agent do with Zendesk data that a Zendesk report cannot?

Zendesk reports what was recorded — statuses, tags, dispositions. The skills read what was actually said, so they can measure repeat contact from the conversation rather than from an agent-set field, and find complaints nobody tagged as one.

MCP for Zendesk

MCP built for Zendesk. Read every ticket, call and chat from Claude, ChatGPT or Gemini — fix routing, clear the backlog, audit the macros answering for you. Comes with 149 skills built in.

Try it now

Works with

Ask Claude to run a skill on Zendesk…

Zendesk MCP server. Several MCP servers read Zendesk tickets, including a supported commercial one. They wrap the ticket API; none of them answers a customer-operations question. Rulebase reads the conversations themselves — connect Zendesk and ask Claude, ChatGPT, Gemini or Cursor a question that needs what was actually said, with 149 customer-operations skills carrying the method.

RoutingFind where Zendesk routing sends work to nobodyFind where support routing is failing — misrouted contacts, reassignment chains, unassigned work, and queues nobody owns.

BacklogSee which open tickets are waiting on you

MacrosAudit the macros that answer for you

TagsClean up the tags your reports rely on

01QA scorecard designBuild a scorecard, size the sample, keep auto-fails separateAll providersScorecards and calibration 02Volume forecastingForecast contacts, and work out how many people you needAll providersDemand and capacity 03Routing auditFind misroutes, ping-pong and queues nobody ownsAll providersRouting and backlog 04Backlog triageFind the tickets in the backlog actually at riskAll providersRouting and backlog 05Bot deflection analysisCheck whether the bot actually reduced contactsAll providersChannels and experience 06Handle time analysisFind where handling time goes before anyone blames a personAll providersRouting and backlog 07Repeat contact and FCRMeasure repeat contact properly, then FCR from itAll providersRouting and backlog 08Cost to serveWork out what a contact really costs, defensiblyAll providersCost and content 09Reviewer calibrationSee whether your reviewers agree, and fix bias and noise apartAll providersScorecards and calibration 10CSAT and NPS surveysPick between CSAT, NPS and CES, and fix response biasAll providersChannels and experience 11AI agent evaluationTest an AI agent before shipping a prompt or model changeAll providersAI oversight 12Customer issue researchName why customers contact you, in causes you can removeAll providersDemand and capacity

Zendesk and MCP

Is there an MCP server for Zendesk?: Yes, several. Swifteq publishes a supported one on the Zendesk Marketplace, and there are community servers on GitHub. All of them wrap the ticket API: fetch a ticket, search tickets, read comments. Zendesk's own MCP support is the client side, letting Zendesk call your tools rather than letting your AI client read its data. Rulebase is a different thing: it syncs the conversations into one normalised record you can sample, join across channels and measure, with 149 customer-operations skills on top.
Do I have to give an AI agent write access to Zendesk?: No. All but one skill in the library is read-only, and the MCP server reads from Rulebase rather than acting on your helpdesk. Anything that would change a record is marked as such and asks before it writes.
What can an AI agent do with Zendesk data that a Zendesk report cannot?: Zendesk reports what was recorded — statuses, tags, dispositions. The skills read what was actually said, so they can measure repeat contact from the conversation rather than from an agent-set field, and find complaints nobody tagged as one.

The other MCP servers for Zendesk

MCP Server for Zendesk
Supportedby Swifteq
Generally available and free, listed on the Zendesk Marketplace. Fetches and reads ticket data on demand from Claude, ChatGPT or Cursor.
zendesk-mcp-server
Communityby reminia
Tickets and comments, plus prompts for ticket analysis and drafting replies.
MCP tools in action flows
From the vendorby Zendesk
The client side: Zendesk calling your tools mid-conversation. It does not expose your tickets to an AI client you choose.

Use one of these if you want to read or update Zendesk records from an AI client — that is what they are for. Rulebase is for the questions they cannot answer: whether the bot is deflecting or merely containing, whether your reviewers agree, whether a complaint was identified, whether first contact resolution is real. Those need the conversation rather than the recorded fields, joined across channels so one customer counts once, in enough volume to sample defensibly.

How to connect Zendesk to Claude, ChatGPT or Gemini

Sync Zendesk into Rulebase. Connect Zendesk to a Rulebase workspace. Rulebase reads a copy of the conversation record; Zendesk stays the system of record.
Add the Rulebase MCP server to your AI client. Point Claude, ChatGPT, Gemini or Cursor at the Rulebase MCP server. The conversations, calls and transcripts become readable from the client you already use.
Install the skills and ask. Run npx rulebase-skills install --all, then ask a question. The matching skill runs against your Zendesk conversations and answers from what was actually said.

Other platforms: MCP for Freshdesk, MCP for Help Scout, MCP for Crisp, MCP for Aircall, MCP for Five9, MCP for Genesys Cloud CX, MCP for NICE CXone, MCP for Avaya. Browse all 149 skills. Last reviewed 2026-08-10 · see which platforms publish an MCP server.

Compliance and security
guardrails built in.

SOC 2 Type II. GDPR. PCI DSS. Zero AI training on your data. Built for the compliance demands of regulated financial services.

SOC Type 2

We meet SOC 2 requirements to ensure secure and compliant management of data across all our systems.

GDPR

With our operations in Ireland, we operate under GDPR – the world's strictest standard for data privacy.

PCI DSS

We protect cardholder and payment data in line with PCI DSS standards, ensuring secure processing across all transactions.

Encryption

We encrypt your data with AES-256 at rest and TLS in transit, ensuring end-to-end protection for every byte.

Review every conversation. Act on what it finds.

AI for customer operations, built for financial services. Specialist agents chase every issue to resolution and every stalled customer to activation.