MCP Server
MCP Server for Zendesk by Swifteq connects Claude, ChatGPT and other MCP-compatible AI assistants to your Zendesk account. Your assistant can search and read tickets, conversations, customers, Help Center articles, CSAT and ticket metrics, and, if you allow it, reply to and create tickets, draft Help Center articles and curate macros. 👉 Start free: create your Swifteq account What your AI assistant can do ✅ Support reports and KPIs from one prompt: first reply time, resolution time, reopen and one-touch rates, by group, brand, agent, channel, week or month. ✅ Reports over months of tickets (Team plan): ticket counts, CSAT, and reply and resolution times by group, agent, tag or month, in one call, from the optional ticket data store. ✅ Find what is driving ticket volume: search tickets with any filter and read full conversations, attachments and screenshots, across email, web, messaging and WhatsApp. ✅ Reply to, update and create tickets: internal notes or public replies, fields, tags, CCs, files and macros, and link incidents to a problem ticket. ✅ Update many tickets at once (Team plan): change a custom field, tags, status or assignee on up to 100 tickets, or rewrite the subject of each, after you confirm. ✅ Keep your Help Center up to date: find content gaps, draft articles in your house style, and update articles, translations, categories and sections across every brand. ✅ Curate your macros: find unused and overlapping macros, then reword, create or deactivate them. ✅ Customers, CSAT and configuration: organizations and users with their custom fields, satisfaction ratings, triggers, fields and forms. Works with your AI assistant Claude (web, desktop and Claude Code) and ChatGPT are supported. Cursor, VS Code with GitHub Copilot and Microsoft Copilot Studio work too, and so does any client that supports MCP with OAuth sign-in. Setup guides Permissions and security 🔒 Each teammate connects their own Zendesk login, so their Zendesk role and groups apply. 🔒 Read-only unless you allow changes. Public replies, publishing articles, changes to users and organizations, changes to macros, and merging tickets each stay off until an account owner turns them on. 🔒 Nothing can be deleted: there are no delete or spam tools, and a merge only happens once an account owner has turned merging on. 🔒 Hosted on AWS in Ireland. Swifteq holds SOC 2 Type 2 certification. Pricing Free for 500 MCP calls a month per Zendesk account, no credit card required. The Team plan has unlimited calls: 49 EUR a month covers your account and its first user, and each extra user is 9 EUR a month, so a team of 10 pays 130 EUR a month. Paid annually, it is 490 EUR a year plus 84 EUR a year per extra user. See pricing About the developer: Swifteq is an independent company in Dublin, Ireland, and a qualified Zendesk technology partner. Our Automation, Help Center and Agent Assist apps help support teams work smarter. Data privacy: The MCP Server reads the Zendesk data your login can see (tickets, users, organizations, Help Center, CSAT and configuration) only when your AI assistant asks, and passes it to the assistant. It never stores ticket conversations or personal information (PII) from Zendesk about your customers or employees. An optional ticket data store, which an account owner switches on for reports, keeps ticket fields, tags and metrics. Swifteq keeps your account email and request logs without ticket content or search text, runs no AI model and does not train on your data. Details: Security and data handling and swifteq.com/data-zendesk-apps
Watari
Customer bugs, located in your code. Watari reads each Zendesk ticket you send it, including screenshots, screen recordings, PDFs, voice notes and logs, and turns it into a structured bug report: severity, steps to reproduce, impact, and the customer's own words. It then ranks the files and functions in your repository most likely to contain the defect, with the reasoning for each. A draft fix your engineers review. When it is confident enough, Watari opens a draft pull request on GitHub (or a pull request on Azure DevOps). Your reviewers and your own CI decide what ships. Watari never merges. Duplicate reports collapse into one bug. When five customers report the same regression, your engineers see one issue, filed in Jira, Linear, GitHub Issues or Azure Boards, with every ticket linked to it. Customers hear back when the fix is live. Watari drafts a plain-language root-cause update. After someone on your team approves it and the deploy is confirmed, it is posted to every Zendesk ticket that reported the bug, as a public reply or an internal note (your choice per update). Updates can be translated into each customer's language. How it works with Zendesk - You choose one or more Zendesk tags. Watari creates a webhook and a trigger in your account that forward tickets carrying any of those tags. Untagged tickets are never sent. - The only thing Watari writes to a ticket is the root-cause update your team approved. It does not change ticket status, tags or assignees. - Disconnecting removes the webhook and trigger it created. Also works with Intercom and Freshdesk for support, GitHub and Azure DevOps for code, Slack and Microsoft Teams for alerts. Pricing. You pay per Mapped Issue: a ticket Watari turns into a bug or feature request and maps to your code with high confidence. Draft pull requests, root-cause updates and issue-tracker sync are included. The 14-day trial includes 10 Mapped Issues and needs no card. Plans at https://watari.ai/pricing. Your data. Integration tokens are encrypted with AES-256-GCM, each workspace is isolated with row-level security, and your data never trains models. Our mapping accuracy, misses included, is public at https://watari.ai/proof. Data Watari accesses, and why - Reads: only tickets carrying your chosen tags: subject, description, public and internal comments, status, tags, attachments, names and emails of the requester, submitter and assignee, and the requester's organization name. - Why: to write the bug report, find the related code, group duplicates, and address the update to the right people. - Writes: only the update your team approves, as a comment on the reporting tickets. - Stores: in your Watari workspace (AWS us-east-1). Processed by the subprocessors at https://watari.ai/legal/subprocessors. Deleting your workspace deletes it after a 30-day grace period. Details: https://watari.ai/legal/privacy