Zendesk MCP Server

Zendesk MCP Server

Bring Zendesk Support & the Help Center into your AI assistant. A Model Context Protocol (MCP) server that lets your assistant search articles, answer questions, and create, track and update tickets in plain language — without switching apps.

Think of it as the Zendesk agent for Microsoft 365 Copilot, but vendor-neutral — it drops into any MCP client (Claude Desktop, Claude Code, Cursor, VS Code, …) instead of being tied to one assistant — and it always acts with each user's own Zendesk permissions, never a shared admin key.

What your assistant can do

Ask in natural language; the assistant figures out context and intent, then calls the right tools on your behalf:

• Find answers in the Help Center — "how do I request a software license?" or "what's the time-off policy?" surfaces the right article, by meaning, not just keywords.
• Create, view and update tickets without leaving the conversation — open a ticket, check its status, add a public reply or an internal note, change the priority or assignee, or mark it solved.
• Summarize a ticket for reporting or a quick decision — pull the details and the full comment thread and get the gist in a sentence.
• Search and triage your queue in plain language — "show me my open tickets about billing from this week."
• Draft and maintain knowledge-base articles — write a new article, or revise a large one one section at a time so the whole HTML body never has to round-trip through the model.

Because it runs on the user's own OAuth session, the assistant only ever sees and touches what that person is allowed to — the same scoping you'd get signing into Zendesk directly.

How it's different

Most Zendesk integrations use a shared admin API key, giving every user full access to every ticket, and bolt on a fixed set of tools. This server is built differently:

• Per-user authentication, OAuth-only — In both transports, auth is OAuth 2.1 PKCE: each user authenticates with their own Zendesk credentials, so the assistant sees exactly what the user is allowed to see. Static API tokens are deliberately not supported (see below).
• Two deployment shapes, same auth story — Run it on your laptop as a stdio MCP server (Claude Desktop / Claude Code / VS Code) or deploy it as a private remote MCP server with one user, one Zendesk session per HTTP request.
• Context-friendly tool modes — Expose every operation as its own tool, group them into namespace proxies, or collapse to a single unified tool. Tools are segmented into namespaces you can selectively enable, so each context loads only the surface it needs.
• Section-based article editing — For large Help Center articles, read and rewrite one section at a time (parsed by h1/h2/h3 headings) instead of shuffling the full HTML body through the assistant. Reduces tokens by 10–100× on targeted edits.
• Read-only mode — Restrict the server to read operations only, ideal for assistants that should never modify data.
• Lean stack — Built on the official @modelcontextprotocol/sdk plus zod.

Under the hood it speaks to the Zendesk Support & Help Center (Guide) APIs, runs locally over stdio or as a private remote MCP server over HTTP, and ships fine-grained tool-visibility controls — the specifics are below.

When to use this server

Reach for it when:

• You want an LLM to read or triage Zendesk tickets and Help Center articles on behalf of a real user, with that user's own permissions — not a shared admin key.
• You're editing large Help Center articles and want section-scoped reads/rewrites instead of round-tripping the full HTML body through the model.
• You need to cap the tool surface — read-only assistants, a single namespace, or one unified tool to fit a tight context budget.
• You run a stdio MCP client (Claude Desktop, Claude Code, Cursor, VS Code, Cline, …) and want a npx-installable server with no extra infrastructure, or you want to deploy it as a private remote MCP server that web/native clients reach over HTTP — each MCP client still carries its own user's OAuth token.

Look elsewhere when:

• You need Zendesk products outside Support & Guide (e.g. Talk, Explore analytics, Sell) — those endpoints aren't covered.
• You need a single shared service account, or static API-token auth — this server doesn't support either, by design (see What this server does not do).

What this server does not do

No API-token authentication. This server is OAuth 2.1 PKCE only — there is no ZENDESK_EMAIL + ZENDESK_API_TOKEN (Basic auth) mode, in any transport. This is a deliberate design choice:

• API tokens are insufficiently secure. A Zendesk API token is a long-lived, static, shared secret that carries the full rights of the issuing user — no per-user scoping, no short expiry, no per-user consent or revocation. OAuth 2.1 PKCE issues per-user, revocable tokens instead, so the LLM only ever sees what the authenticated user is allowed to see.
• API tokens don't scale. A single static credential can't attribute actions to individual users or be revoked granularly, and it makes a multi-user remote deployment unsafe (in HTTP it would expose the issuing user's rights to every caller). OAuth scales naturally: each MCP client carries its own user's token.

If you specifically need an API-token / service-account mode (e.g. headless CI with a shared account), use one of the other Zendesk MCP servers that support it — see Inspiration & related projects.

Use cases

Tool modes

The server registers tools in one of three modes, controlled by --mode:

In namespace and single modes, the proxy tool accepts { "operation": "<tool_name>", "params": { ... } } and dispatches to the appropriate handler after validating params through the original Zod schema. Proxy descriptions include only the first sentence of each sub-operation to stay compact; the full schema is applied when the operation is actually called.

Tip: The single mode is particularly useful for models with limited tool slots — one tool handles every operation.

Scoping the surface

--namespace and --read-only apply to every mode (including the default namespace mode) — they filter tools before the proxies are built, so the description of each proxy reflects only the operations that survive the filters. Combine them to register a focused surface:

# Only the Help Center proxy, only read-only operations
zendesk-mcp-server acme --namespace help_center --read-only

# Only the Tickets proxy (read + write)
zendesk-mcp-server acme --namespace tickets

--namespace is repeatable. --tool is also available for cherry-picking individual operations but forces --mode all.

Available tools

Help Center context (instructions + resources)

Beyond tools, the server hands an LLM the structural context it needs to work against your Help Center — so it stops guessing locales or fuzzy-matching section names and uses real IDs instead. This is delivered through two MCP-native channels (both active only when the help_center namespace is, and disabled together with --no-topology):

• instructions (sent on initialize): a short, static blob auto-loaded by compliant clients. It names the subdomain and points at the topology resource.
• zendesk-hc://topology (a pull-only MCP resource): read on demand, it returns Markdown describing the active locales (and the default), the category → section tree with IDs, the visibility user segments, the permission groups, and the calling user's role. It is fetched with the caller's own token, so it respects that user's read permissions. On a very large Help Center the section tree is summarized (per-category, with a pointer to list_sections) to stay concise.

Clients that don't consume instructions or resources simply ignore them — the feature degrades silently. Use --no-topology to turn both off server-wide.

Prerequisites

• Node.js >= 20 (runtime — declared in package.json#engines.node)
• A Zendesk instance (Support or Suite)

Contributors and maintainers run the toolchain on a newer Node + pnpm — see Development.

Quick start: local (stdio)

The default mode. One developer, one Zendesk account, OAuth 2.1 PKCE in the browser.

Install

# Run without installing
npx -y @fruggr/zendesk-mcp-server <your-subdomain>

# Or install globally
npm install -g @fruggr/zendesk-mcp-server
zendesk-mcp-server <your-subdomain>

Cloning from source and running a development branch is covered in the Development section.

Zendesk OAuth setup

1. Go to Admin Center → Apps and integrations → APIs → OAuth Clients
2. Create a public client:
   • Identifier: <your-subdomain>_zendesk (or set ZENDESK_OAUTH_CLIENT_ID)
   • Redirect URL: http://localhost:27439/callback (change the port to match ZENDESK_OAUTH_CALLBACK_PORT / --callback-port if you override it; Zendesk accepts several redirect URLs, one per line)

Run

zendesk-mcp-server <your-subdomain>

On the first tool call, the server starts the sign-in flow: it opens a browser window and returns a tool message containing the authorize URL. The call does not block waiting for sign-in — authenticate in the browser (or open the URL manually if it didn't open), then retry the request.

Once authenticated, the token is persisted to disk (one owner-only 0600 file per subdomain in your OS config dir — %APPDATA%\fruggr\zendesk-mcp-server\<subdomain>.json on Windows, ${XDG_CONFIG_HOME:-~/.config}/fruggr/zendesk-mcp-server/<subdomain>.json elsewhere; override the path with ZENDESK_TOKEN_FILE). It is reused across restarts, so you don't re-authenticate every time the MCP client respawns the server. If the Zendesk OAuth client has token expiration enabled, the stored refresh token is used to renew access silently — proactively (the token is refreshed before use when it's expired, near expiry, or of unknown age, so the first request after an overnight gap never hits a visible auth error) and periodically in the background so a long-lived, idle session never serves a stale token. Only an expired/invalid refresh token triggers a new browser sign-in.

Port conflict? If port 27439 is already in use the first tool call returns a clear error telling you to set ZENDESK_OAUTH_CALLBACK_PORT (or --callback-port) to a free port — remember to register the matching http://localhost:<port>/callback redirect URL in your Zendesk OAuth client.

MCP client wiring

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "zendesk": {
      "command": "npx",
      "args": ["-y", "@fruggr/zendesk-mcp-server", "<your-subdomain>", "--mode", "single"]
    }
  }
}

claude mcp add zendesk -- npx -y @fruggr/zendesk-mcp-server <your-subdomain> --mode single

Add to your .vscode/mcp.json:

{
  "servers": {
    "zendesk": {
      "command": "npx",
      "args": ["-y", "@fruggr/zendesk-mcp-server", "<your-subdomain>", "--mode", "single"]
    }
  }
}

Quick start: remote (HTTP)

🧪 Experimental. The HTTP transport is shipped but has not yet been exercised end-to-end against a real Zendesk tenant from every supported MCP client. Local stdio is the supported path. Until this notice is removed, expect rough edges around OAuth discovery behind reverse proxies, CORS with browser clients, and 401 / refresh flows — please open an issue with the symptoms you hit.

Deploy a private MCP server for one Zendesk account. Every MCP client connecting to the server presents its own user's OAuth bearer in Authorization: — the server never sees a shared admin key.

Zendesk OAuth setup

Same procedure as the local quick start, with one difference: the Redirect URL must match the callback your MCP client uses — provided by the client itself, e.g. https://claude.ai/oauth/callback for claude.ai on the web. Check your client's docs.

Run the server

zendesk-mcp-server <your-subdomain> --transport http --port 3000 \
  --public-url https://mcp.example.com
# stderr: Zendesk MCP server running via http on 0.0.0.0:3000

Public URL

--public-url (or PUBLIC_URL=…) is the URL clients use to reach you. It's what gets advertised in the OAuth discovery metadata as the canonical resource identifier (RFC 8707). When the server is behind a TLS reverse proxy — Azure App Service, Heroku, Fly.io, Cloudflare Tunnel, nginx, Caddy… — the bind host and the public URL differ, and spec-compliant MCP clients will refuse the connection if the metadata advertises the wrong resource. Without it the server boots in a degraded mode and prints a warning.

Authentication on every request

Authorization: Bearer … is required on every /mcp request — a session id alone is never accepted as a credential. The most recent bearer presented on a session is the one used for Zendesk calls, so a client refreshing its token mid-session just works.

Verify discovery endpoints

Served by the HTTP transport in src/transports/http.ts:

curl -s http://localhost:3000/.well-known/oauth-protected-resource
# → { "authorization_servers": ["https://<subdomain>.zendesk.com"], ... }

curl -s http://localhost:3000/.well-known/oauth-authorization-server
# → { "issuer": "https://<subdomain>.zendesk.com", "authorization_endpoint": "...", ... }

curl -s -i http://localhost:3000/healthz   # → 200 OK

MCP client wiring

Every major MCP client supports remote servers over Streamable HTTP and handles the OAuth 2.1 PKCE discovery flow natively — paste the URL, sign in once, you're connected. Replace https://mcp.example.com below with your deployed origin.

claude mcp add zendesk --transport http https://mcp.example.com/mcp

Settings → Connectors → + Add custom connector, paste https://mcp.example.com/mcp, click Connect. Claude Desktop drives the OAuth flow in your browser on first call.

Settings → Connectors → Add custom connector, same URL. The OAuth flow runs in the same tab.

Add to your .vscode/mcp.json:

{
  "servers": {
    "zendesk": {
      "type": "http",
      "url": "https://mcp.example.com/mcp"
    }
  }
}

Both expose an MCP settings UI that accepts a remote URL. Paste https://mcp.example.com/mcp and sign in when prompted.

Zed added native OAuth 2.0 + PKCE for Streamable HTTP MCP servers in 2026 (zed-industries/zed#51768). Configure the remote server in your Zed settings; on first use Zed opens a loopback browser callback to complete the flow.

If you're on an older Zed build that predates that change, fall back to mcp-remote as a local shim that does the OAuth flow on your machine and proxies the session:

{
  "context_servers": {
    "zendesk": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.example.com/mcp"]
    }
  }
}

On the first call the MCP client fetches the discovery metadata, performs the OAuth 2.1 PKCE flow against Zendesk on behalf of the end user, and sends the resulting access token as a Bearer to the server. Each subsequent tool call runs with that user's Zendesk permissions.

CORS

The HTTP transport ships a default CORS allowlist that covers today's major browser-based MCP clients out of the box (ordered by user base): chatgpt.com, claude.ai, gemini.google.com, copilot.microsoft.com, perplexity.ai, chat.mistral.ai, grok.com, plus chat.openai.com. Localhost on any port (MCP Inspector, dev pages) is also always allowed.

Native MCP clients (Claude Desktop / Claude Code CLI / Cursor / VS Code / Zed) send no Origin header — CORS doesn't apply to them, they work regardless.

To allow an additional browser origin (custom dashboard, internal portal), pass --cors-origin (repeatable) or set CORS_ORIGIN as a comma-separated list:

zendesk-mcp-server acme --transport http --port 3000 \
  --cors-origin https://internal-dashboard.example.com \
  --cors-origin https://team-portal.example.com

The defaults are always applied — your additions extend them, they don't replace them.

Operator responsibilities

This server provides the MCP transport and the OAuth discovery metadata. The operator is still responsible for:

• TLS termination (put the server behind a reverse proxy like Caddy / nginx / Cloudflare Tunnel)
• Network exposure & firewall (the server binds 0.0.0.0 by default — choose carefully)
• Process supervision (systemd, Docker, fly.io, your hosting provider's runner — none is shipped here)

CLI reference

zendesk-mcp-server <subdomain> [options]

Options:
  --mode <mode>           single | namespace (default) | all
  --namespace <ns>        Filter by namespace (repeatable): tickets, help_center, users
  --tool <name>           Filter by tool name (repeatable, forces --mode all)
  --read-only             Only expose read operations
  --no-topology           Disable the Help Center structural context
                          (instructions + zendesk-hc://topology resource)
  --log-level <level>     debug | info (default) | warn | error
  --transport <t>         stdio (default) | http
  --host <host>           HTTP bind host (default: 0.0.0.0)
  --port <port>           HTTP bind port (default: 3000; 0 = OS-assigned)
  --public-url <url>      Public URL clients use to reach the server (HTTP mode,
                          required behind a TLS reverse proxy)
  --cors-origin <url>     Extra browser origin allowed by CORS (repeatable;
                          adds to the default allowlist of major web MCP
                          clients + localhost-any-port)
  --callback-port <port>  Local OAuth callback port for stdio (default 27439)

--namespace and --read-only are applied before the proxies are registered, so they narrow the surface in every mode — in the default namespace mode, --namespace help_center registers a single proxy (zendesk_help_center) instead of three.

Examples:

# Local single-tool mode — minimal context, every operation in one tool
zendesk-mcp-server acme --mode single

# Read-only tickets only
zendesk-mcp-server acme --read-only --namespace tickets

# Cherry-pick specific tools
zendesk-mcp-server acme --tool get_ticket --tool search_tickets --tool get_current_user

# Remote HTTP, read-only Help Center surface
zendesk-mcp-server acme --transport http --port 8080 \
  --namespace help_center --read-only

Environment variables

The server uses per-user OAuth 2.1 PKCE for every transport (local stdio and remote HTTP). There is no static API-token mode — see What this server does not do.

Troubleshooting

The browser doesn't open during OAuth login

The OAuth flow opens your default browser on the first tool call. The first call fails fast with a message that includes the authorize URL, so even if the browser can't open (common in sandboxed or remote desktop environments) you can open that URL manually — it's also printed to the server's stderr. Sign in, then retry the request.

To collect diagnostics, restart with LOG_LEVEL=debug. The server then emits structured logs through two channels, so they're reachable on any MCP client:

• stderr — captured to a log file by every mainstream client.
• MCP logging notifications (notifications/message) — surfaced by clients that support the logging capability.

When the browser fails to open, look for the oauth_browser_open_failed event: it reports the underlying error, the platform, and which environment markers are present (no secrets, tokens, or env values are ever logged).

The OAuth callback port is already in use

The sign-in flow runs a short-lived local server on port 27439 to receive the callback. If that port is taken, the first tool call fails with a message saying so (and logs oauth_callback_listen_failed). Pick a free port with ZENDESK_OAUTH_CALLBACK_PORT=<port> (or --callback-port <port>), and register the matching http://localhost:<port>/callback redirect URL in your Zendesk OAuth client.

I have to re-authenticate every time

The OAuth token is persisted to an owner-only file in your OS config dir and reused across restarts, so this shouldn't happen. If it does, check that the file is writable (ZENDESK_TOKEN_FILE to relocate it) and look for token_persist_failed in the logs.

Where each client writes the server's stderr:

Development

Toolchain

The toolchain (Node 24 + pnpm 11) is used to build, lint, type-check and test the project. The published package still runs on Node 20+ (see engines.node); a dedicated CI job installs the packed tarball on Node 20 and runs the smoke test to keep that promise honest.

# Clone, install, build
git clone https://github.com/fruggr/zendesk-mcp-server.git
cd zendesk-mcp-server && pnpm install && pnpm build
node dist/index.js <your-subdomain>

# Dev mode, OAuth (browser opens on first tool call)
pnpm dev -- <your-subdomain> --mode all

# Dev mode, HTTP transport (OAuth bearer from the MCP client)
pnpm dev -- <your-subdomain> --transport http --port 3000 --public-url http://localhost:3000

# Build / typecheck / lint / test
pnpm build && pnpm typecheck && pnpm check && pnpm test

To test a PR branch without publishing to npm — the prepare script builds on install:

npx -y github:fruggr/zendesk-mcp-server <your-subdomain>
npx -y github:fruggr/zendesk-mcp-server#my-feature-branch <your-subdomain>

Contributor conventions (architecture, code style, submission bar, release workflow) live in AGENTS.md.

Inspiration & related projects

This project was built with reference to:

• The official Zendesk API documentation
• mattcoatsworth/zendesk-mcp-server
• koundinya/zd-mcp-server

Releases & versioning

Versions follow SemVer and are calculated automatically from commit messages — no one bumps the version by hand. Every merge to main triggers semantic-release, which inspects the new Conventional Commits since the previous tag, computes the next version, updates CHANGELOG.md, publishes to npm, creates the matching GitHub Release, and mirrors the release into the official MCP registry as io.github.fruggr/zendesk-mcp-server so registry-driven clients discover the new version automatically.

FAQ

Do I need a Zendesk admin API key? No — and the server doesn't support one. The OAuth 2.1 PKCE flow means each user authenticates with their own credentials and the server acts with exactly their permissions. Static API tokens are intentionally unsupported (see What this server does not do).

Which Zendesk products are supported? Zendesk Support (tickets, users, organizations) and the Help Center / Guide (articles, sections, categories, translations, labels, content tags, segments, attachments). Talk, Explore, and Sell are out of scope.

How do I keep the model's context small? Use --mode single (one zendesk tool) or --mode namespace (three proxies), and --read-only to drop write operations. For big articles, the section-based tools (get_article_outline, get_article_section, update_article_section) let the model touch one section at a time instead of the whole HTML body.

Can I restrict it to read-only? Yes — pass --read-only and every write tool is filtered out before the proxies are built, in any mode.

Which Node.js version do I need? Node.js >= 20 to run the published package (engines.node). The dev toolchain uses a newer Node — see Development.

The OAuth browser window didn't open. What now? The authorization URL is also printed to stderr — open it manually. Restart with LOG_LEVEL=debug for the full flow trace. See Troubleshooting.

Is it safe to run via npx? Releases are published from CI via npm Trusted Publishing (OIDC), so each version carries a build provenance attestation you can verify on its npm page. No secrets are ever logged by the server.

Contributing

Pull requests are welcome — including AI-assisted ones, as long as the human author has read and validated every line.

The full guide is in CONTRIBUTING.md. The short version:

1. Fork and create a feature branch from main.
2. Practice TDD: write the failing test first, then implement.
3. Use Conventional Commits — they drive the next version bump via semantic-release.
4. Make pnpm check, pnpm typecheck, and pnpm test pass locally.
5. Run a Claude Code review on your diff before pushing.
6. Open a PR.

Every PR is reviewed automatically by CodeRabbit in CI, on top of the author-side AI review. The project is maintained in part with Claude Code assistance; that workflow is documented in CONTRIBUTING.md.

License

MIT

---

Built and maintained by Digital4better for the Fruggr project.