@lincolnminto/zendesk-mcp-server 2.20.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +21 -0
- package/README.md +309 -0
- package/dist/index.js +5144 -0
- package/glama.json +4 -0
- package/package.json +116 -0
package/LICENSE
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2026 Digital4Better
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
package/README.md
ADDED
|
@@ -0,0 +1,309 @@
|
|
|
1
|
+
# Zendesk MCP Server
|
|
2
|
+
|
|
3
|
+
[](https://www.npmjs.com/package/@lincolnminto/zendesk-mcp-server)
|
|
4
|
+
[](LICENSE)
|
|
5
|
+
[](https://nodejs.org)
|
|
6
|
+
|
|
7
|
+
A [Model Context Protocol](https://modelcontextprotocol.io) (MCP) server that
|
|
8
|
+
puts Zendesk inside your AI assistant. It finds answers in the Help Center;
|
|
9
|
+
drafts, updates and translates articles while keeping the languages in sync; and
|
|
10
|
+
handles Support tickets end to end, comments, triage and image attachments
|
|
11
|
+
included. It all happens in plain language, without switching apps.
|
|
12
|
+
|
|
13
|
+
It does roughly what the
|
|
14
|
+
[Zendesk agent for Microsoft 365 Copilot](https://support.zendesk.com/hc/en-us/articles/9958331458458-Using-the-Zendesk-agent-in-Microsoft-365-Copilot)
|
|
15
|
+
does, minus the tie to one vendor: it drops into any MCP client (Claude Desktop,
|
|
16
|
+
Claude Code, Cursor, VS Code, and the rest). And it always acts with each user's
|
|
17
|
+
own Zendesk permissions, never a shared admin key.
|
|
18
|
+
|
|
19
|
+
## What your assistant can do
|
|
20
|
+
|
|
21
|
+
Ask in natural language; the assistant works out the context and the intent,
|
|
22
|
+
then calls the right tools on your behalf.
|
|
23
|
+
|
|
24
|
+
- Find answers in the Help Center. "How do I request a software license?" or
|
|
25
|
+
"what's the time-off policy?" surfaces the right article, matched by meaning
|
|
26
|
+
rather than by keyword.
|
|
27
|
+
- Create, view and update tickets without leaving the conversation: open a
|
|
28
|
+
ticket, check its status, add a public reply or an internal note, change the
|
|
29
|
+
priority or the assignee, mark it solved.
|
|
30
|
+
- Summarize a ticket for a report or a quick decision. The assistant pulls the
|
|
31
|
+
details and the full comment thread and gives you the gist in a sentence.
|
|
32
|
+
- Read the screenshots and photos attached to a ticket. Error dialogs, UI
|
|
33
|
+
captures and product photos are handed to your assistant's own model as
|
|
34
|
+
images, so it can describe them or act on what they show.
|
|
35
|
+
- Search and triage your queue in plain language: "show me my open tickets about
|
|
36
|
+
billing from this week."
|
|
37
|
+
- Draft and maintain knowledge-base articles. You can write a new one, or revise
|
|
38
|
+
a large one a single section at a time, so the whole HTML body never has to
|
|
39
|
+
round-trip through the model.
|
|
40
|
+
|
|
41
|
+
## Why this server
|
|
42
|
+
|
|
43
|
+
Most Zendesk integrations run on a shared admin API key, which hands every user
|
|
44
|
+
full access to every ticket, and bolt on a fixed set of tools. This one is built
|
|
45
|
+
differently.
|
|
46
|
+
|
|
47
|
+
- Per-user authentication by default. Both transports default to OAuth 2.1
|
|
48
|
+
PKCE: each user signs in with their own Zendesk credentials, so the assistant
|
|
49
|
+
sees and touches exactly what that person is allowed to, the same scoping you
|
|
50
|
+
get by signing into Zendesk directly. A static API-token escape hatch is
|
|
51
|
+
available for stdio-only CI / headless contexts; HTTP accepts only OAuth
|
|
52
|
+
([why](#what-this-server-does-not-do)).
|
|
53
|
+
- Section-based article editing. For large Help Center articles, read and
|
|
54
|
+
rewrite one section at a time (parsed by `h1`/`h2`/`h3` headings) instead of
|
|
55
|
+
shuffling the full HTML body through the assistant. On a targeted edit that
|
|
56
|
+
cuts tokens by a factor of 10 to 100.
|
|
57
|
+
- Native multimodal attachments. Ticket images come back as native MCP image
|
|
58
|
+
content, so the client's own model (Claude, GPT, Gemini, whichever) sees the
|
|
59
|
+
pixels directly. No server-side vision model, no extra API key, and nothing
|
|
60
|
+
tying you to one provider. Non-image attachments come back as text references,
|
|
61
|
+
and both image caps are configurable.
|
|
62
|
+
- A tool surface you can cap. Expose every operation as its own tool, group them
|
|
63
|
+
into namespace proxies, or collapse everything into a single unified tool. You
|
|
64
|
+
can also filter by namespace or down to read-only operations, so each context
|
|
65
|
+
loads only the surface it needs (see [Tool surface](#tool-surface)).
|
|
66
|
+
- Two deployment shapes, same auth story. Run it on your laptop as a stdio MCP
|
|
67
|
+
server, or deploy it as a private remote MCP server reached over HTTP, with one
|
|
68
|
+
Zendesk session per request and each client carrying its own user's token.
|
|
69
|
+
- A lean stack: the official `@modelcontextprotocol/sdk` plus `zod`, speaking to
|
|
70
|
+
the Zendesk Support and Help Center (Guide) APIs.
|
|
71
|
+
|
|
72
|
+
Look elsewhere when:
|
|
73
|
+
|
|
74
|
+
- You need Zendesk products outside Support and Guide (Talk, Explore analytics,
|
|
75
|
+
Sell). Those endpoints aren't covered.
|
|
76
|
+
- You need a single shared service account for a multi-user or remote
|
|
77
|
+
deployment. This server refuses that by design (the API-token escape hatch is
|
|
78
|
+
stdio-only CI, not a way to turn this into a shared backend — see below).
|
|
79
|
+
|
|
80
|
+
## What this server does *not* do
|
|
81
|
+
|
|
82
|
+
There is no shared or HTTP-wide API-token authentication. Both transports
|
|
83
|
+
default to OAuth 2.1 PKCE, and HTTP accepts *only* that — a static API token is
|
|
84
|
+
refused at boot there. That is deliberate, for two reasons.
|
|
85
|
+
|
|
86
|
+
1. **API tokens are insufficiently secure.** A Zendesk API token is a
|
|
87
|
+
long-lived, static, shared secret that carries the full rights of the user
|
|
88
|
+
who issued it. There is no per-user scoping, no short expiry, and no per-user
|
|
89
|
+
consent or revocation. OAuth 2.1 PKCE issues per-user, revocable tokens
|
|
90
|
+
instead.
|
|
91
|
+
2. **API tokens don't scale.** A single static credential can't attribute
|
|
92
|
+
actions to individual users, and it can't be revoked granularly. It also
|
|
93
|
+
makes a multi-user remote deployment unsafe: over HTTP it would expose the
|
|
94
|
+
issuing user's rights to every caller.
|
|
95
|
+
|
|
96
|
+
The one exception is **stdio, for headless/CI use**: set `ZENDESK_EMAIL` +
|
|
97
|
+
`ZENDESK_API_TOKEN` and the server uses Basic auth instead of starting the
|
|
98
|
+
OAuth browser flow — see [`docs/api-token-stdio.md`](docs/api-token-stdio.md).
|
|
99
|
+
Reason 2 still applies there in full; reason 1's "every caller" clause doesn't,
|
|
100
|
+
since a stdio server is one local process for one operator, with no second
|
|
101
|
+
caller to expose rights to.
|
|
102
|
+
|
|
103
|
+
If you need a *shared* service account, or static API-token auth over HTTP,
|
|
104
|
+
use one of the other Zendesk MCP servers that support it. A few are listed
|
|
105
|
+
under [Inspiration & related projects](#inspiration--related-projects).
|
|
106
|
+
|
|
107
|
+
## Quick start: local (stdio)
|
|
108
|
+
|
|
109
|
+
The default shape: one developer, one Zendesk account, OAuth 2.1 PKCE in the
|
|
110
|
+
browser. You need **Node.js >= 20** and a **Zendesk** instance (Support or
|
|
111
|
+
Suite).
|
|
112
|
+
|
|
113
|
+
### Install
|
|
114
|
+
|
|
115
|
+
```bash
|
|
116
|
+
# Run without installing
|
|
117
|
+
npx -y @lincolnminto/zendesk-mcp-server <your-subdomain>
|
|
118
|
+
|
|
119
|
+
# Or install globally
|
|
120
|
+
npm install -g @lincolnminto/zendesk-mcp-server
|
|
121
|
+
zendesk-mcp-server <your-subdomain>
|
|
122
|
+
```
|
|
123
|
+
|
|
124
|
+
Signing in needs a Zendesk OAuth client, so register one first (next section).
|
|
125
|
+
|
|
126
|
+
### Zendesk OAuth setup
|
|
127
|
+
|
|
128
|
+
1. Go to **Admin Center → Apps and integrations → APIs → OAuth Clients**
|
|
129
|
+
2. Create a **public** client:
|
|
130
|
+
- **Identifier**: `<your-subdomain>_zendesk` (or set `ZENDESK_OAUTH_CLIENT_ID`)
|
|
131
|
+
- **Redirect URL**: `http://localhost:27439/callback` (change the port to match
|
|
132
|
+
`ZENDESK_OAUTH_CALLBACK_PORT` / `--callback-port` if you override it; Zendesk
|
|
133
|
+
accepts several redirect URLs, one per line)
|
|
134
|
+
|
|
135
|
+
On the first tool call the server starts the sign-in flow: it opens a browser
|
|
136
|
+
window and returns the authorize URL in a tool message. The call does not block
|
|
137
|
+
waiting for sign-in, so authenticate in the browser and then retry the request.
|
|
138
|
+
The token is persisted to an owner-only file and reused across restarts, so you
|
|
139
|
+
don't authenticate again every time your MCP client respawns the server (path
|
|
140
|
+
and overrides: [`ZENDESK_TOKEN_FILE`](docs/configuration.md#zendesk_token_file)).
|
|
141
|
+
|
|
142
|
+
> **API token escape hatch (stdio only).** For headless/CI environments where a
|
|
143
|
+
> browser is unavailable, set `ZENDESK_EMAIL` + `ZENDESK_API_TOKEN` (generate
|
|
144
|
+
> the token in **Admin Center → Apps and integrations → APIs → Zendesk API →
|
|
145
|
+
> Token Access**). The server then uses Basic auth instead of starting the
|
|
146
|
+
> OAuth flow. This mode is **refused at boot in HTTP** because a shared static
|
|
147
|
+
> credential would expose every caller to the issuing user's rights. Full setup:
|
|
148
|
+
> [`docs/api-token-stdio.md`](docs/api-token-stdio.md).
|
|
149
|
+
|
|
150
|
+
### MCP client wiring
|
|
151
|
+
|
|
152
|
+
<details>
|
|
153
|
+
<summary><strong>Claude Desktop</strong></summary>
|
|
154
|
+
|
|
155
|
+
Add to your `claude_desktop_config.json`:
|
|
156
|
+
|
|
157
|
+
```json
|
|
158
|
+
{
|
|
159
|
+
"mcpServers": {
|
|
160
|
+
"zendesk": {
|
|
161
|
+
"command": "npx",
|
|
162
|
+
"args": ["-y", "@lincolnminto/zendesk-mcp-server", "<your-subdomain>", "--mode", "single"]
|
|
163
|
+
}
|
|
164
|
+
}
|
|
165
|
+
}
|
|
166
|
+
```
|
|
167
|
+
|
|
168
|
+
</details>
|
|
169
|
+
|
|
170
|
+
<details>
|
|
171
|
+
<summary><strong>Claude Code</strong></summary>
|
|
172
|
+
|
|
173
|
+
```bash
|
|
174
|
+
claude mcp add zendesk -- npx -y @lincolnminto/zendesk-mcp-server <your-subdomain> --mode single
|
|
175
|
+
```
|
|
176
|
+
|
|
177
|
+
</details>
|
|
178
|
+
|
|
179
|
+
<details>
|
|
180
|
+
<summary><strong>VS Code (Copilot / Continue / Cline)</strong></summary>
|
|
181
|
+
|
|
182
|
+
Add to your `.vscode/mcp.json`:
|
|
183
|
+
|
|
184
|
+
```json
|
|
185
|
+
{
|
|
186
|
+
"servers": {
|
|
187
|
+
"zendesk": {
|
|
188
|
+
"command": "npx",
|
|
189
|
+
"args": ["-y", "@lincolnminto/zendesk-mcp-server", "<your-subdomain>", "--mode", "single"]
|
|
190
|
+
}
|
|
191
|
+
}
|
|
192
|
+
}
|
|
193
|
+
```
|
|
194
|
+
|
|
195
|
+
</details>
|
|
196
|
+
|
|
197
|
+
Something not working? See [Troubleshooting](docs/troubleshooting.md).
|
|
198
|
+
|
|
199
|
+
## Quick start: remote (HTTP)
|
|
200
|
+
|
|
201
|
+
> **Experimental.** The HTTP transport ships, but it has not yet been exercised
|
|
202
|
+
> end-to-end against a real Zendesk tenant from every MCP client. Local stdio is
|
|
203
|
+
> the supported path.
|
|
204
|
+
|
|
205
|
+
You can also deploy a private remote MCP server for **one** Zendesk account,
|
|
206
|
+
where every MCP client presents its **own** user's OAuth bearer in
|
|
207
|
+
`Authorization:` and the server never sees a shared admin key. The full guide
|
|
208
|
+
covers OAuth setup, `--public-url` behind a reverse proxy, per-platform config,
|
|
209
|
+
the discovery endpoints, MCP client wiring, CORS and what stays the operator's
|
|
210
|
+
job: **[docs/http-deployment.md](docs/http-deployment.md)**.
|
|
211
|
+
|
|
212
|
+
## Tool surface
|
|
213
|
+
|
|
214
|
+
Tools are grouped into four namespaces: **Tickets**, **Help Center**, **Users &
|
|
215
|
+
Organizations** and **Search**. The server registers them in one of three modes,
|
|
216
|
+
so you can trade granularity against context budget:
|
|
217
|
+
|
|
218
|
+
- **`all`**: every operation as its own tool, for clients with good tool selection;
|
|
219
|
+
- **`namespace`** (default): one proxy tool per namespace, a balanced middle ground;
|
|
220
|
+
- **`single`**: a single `zendesk` tool, for models with limited tool slots.
|
|
221
|
+
|
|
222
|
+
Proxies take `{ "operation": "<tool_name>", "params": { … } }` and validate
|
|
223
|
+
`params` through the original schema. `--namespace`, `--tool` and `--read-only`
|
|
224
|
+
filter tools *before* the proxies are built, so each proxy describes only the
|
|
225
|
+
operations that survive.
|
|
226
|
+
|
|
227
|
+
Every tool with its description and its `read`/`write` mode:
|
|
228
|
+
**[docs/mcp-tools-reference.md](docs/mcp-tools-reference.md)**. The flags and
|
|
229
|
+
worked examples: **[docs/configuration.md](docs/configuration.md)**.
|
|
230
|
+
|
|
231
|
+
## Help Center context
|
|
232
|
+
|
|
233
|
+
Beyond tools, the server hands the LLM the structure of *your* Help Center: the
|
|
234
|
+
active locales, the category → section tree with IDs, the visibility segments
|
|
235
|
+
and the permission groups. With those in hand it uses real IDs instead of
|
|
236
|
+
guessing or fuzzy-matching names. It all arrives through MCP-native channels,
|
|
237
|
+
namely the `instructions` blob sent on `initialize` plus pull-only resources for
|
|
238
|
+
the topology and for reading (or pinning) individual articles. The resources are
|
|
239
|
+
fetched with the caller's own token, and clients that don't support resources
|
|
240
|
+
ignore them silently.
|
|
241
|
+
|
|
242
|
+
What's exposed, what the promoted-article pre-listing costs in requests, and how
|
|
243
|
+
to turn each piece off: **[docs/help-center-context.md](docs/help-center-context.md)**.
|
|
244
|
+
|
|
245
|
+
## Configuration
|
|
246
|
+
|
|
247
|
+
The complete reference for the CLI flags (`--mode`, `--namespace`,
|
|
248
|
+
`--read-only`, `--transport`, `--public-url`, and so on) and the environment
|
|
249
|
+
variables (`ZENDESK_SUBDOMAIN`, `ZENDESK_TOKEN_FILE`, `PUBLIC_URL`, the
|
|
250
|
+
attachment-vision caps) lives in
|
|
251
|
+
**[docs/configuration.md](docs/configuration.md)**. Every variable has its own
|
|
252
|
+
anchor, so you can deep-link a specific setting.
|
|
253
|
+
|
|
254
|
+
## Troubleshooting
|
|
255
|
+
|
|
256
|
+
Browser not opening during OAuth login, the callback port already in use, having
|
|
257
|
+
to re-authenticate every time, and `Permission denied` on the Guide-admin
|
|
258
|
+
endpoints are covered in **[docs/troubleshooting.md](docs/troubleshooting.md)**.
|
|
259
|
+
Restart with `LOG_LEVEL=debug` for the full OAuth flow trace.
|
|
260
|
+
|
|
261
|
+
## Development
|
|
262
|
+
|
|
263
|
+
Setting up the repo, the toolchain, dev mode and how to test a PR branch are
|
|
264
|
+
covered in **[CONTRIBUTING.md](CONTRIBUTING.md#development-setup)**. Architecture
|
|
265
|
+
and code-style conventions live in [`AGENTS.md`](AGENTS.md).
|
|
266
|
+
|
|
267
|
+
## FAQ
|
|
268
|
+
|
|
269
|
+
**Which Zendesk products are supported?**
|
|
270
|
+
Zendesk Support (tickets, users, organizations) and the Help Center / Guide
|
|
271
|
+
(articles, sections, categories, translations, labels, content tags, segments,
|
|
272
|
+
attachments). Talk, Explore and Sell are out of scope.
|
|
273
|
+
|
|
274
|
+
**Do I need a Zendesk admin API key?**
|
|
275
|
+
No. Each user authenticates with their own OAuth credentials by default, and
|
|
276
|
+
the server acts with exactly their permissions. A static API token is only a
|
|
277
|
+
stdio-only CI escape hatch, never a shared or HTTP-wide credential
|
|
278
|
+
([why](#what-this-server-does-not-do)).
|
|
279
|
+
|
|
280
|
+
**Is it safe to run via `npx`?**
|
|
281
|
+
Releases are published from CI via npm Trusted Publishing (OIDC), so each version
|
|
282
|
+
carries a build provenance attestation you can verify on its
|
|
283
|
+
[npm page](https://www.npmjs.com/package/@lincolnminto/zendesk-mcp-server). No secrets
|
|
284
|
+
are ever logged by the server.
|
|
285
|
+
|
|
286
|
+
## Contributing
|
|
287
|
+
|
|
288
|
+
Pull requests are welcome, AI-assisted ones included, as long as the human
|
|
289
|
+
author has read and validated every line. The guide, the author checklist and the
|
|
290
|
+
review workflow are in **[CONTRIBUTING.md](CONTRIBUTING.md)**.
|
|
291
|
+
|
|
292
|
+
Versions follow [SemVer](https://semver.org/) and are released automatically from
|
|
293
|
+
[Conventional Commits](https://www.conventionalcommits.org/); the history is in
|
|
294
|
+
[`CHANGELOG.md`](CHANGELOG.md).
|
|
295
|
+
|
|
296
|
+
## Inspiration & related projects
|
|
297
|
+
|
|
298
|
+
This project was built with reference to:
|
|
299
|
+
- The official [Zendesk API documentation](https://developer.zendesk.com/api-reference/)
|
|
300
|
+
- [mattcoatsworth/zendesk-mcp-server](https://github.com/mattcoatsworth/zendesk-mcp-server)
|
|
301
|
+
- [koundinya/zd-mcp-server](https://github.com/koundinya/zd-mcp-server)
|
|
302
|
+
|
|
303
|
+
## License
|
|
304
|
+
|
|
305
|
+
[MIT](LICENSE)
|
|
306
|
+
|
|
307
|
+
---
|
|
308
|
+
|
|
309
|
+
> Built and maintained by [Digital4better](https://digital4better.com) for the [Fruggr](https://www.fruggr.io) project.
|