# SuperOps [Health: Active]

**Category:** 🎧 Support & Service Management  
**Repository:** https://github.com/WYRE-AI/superops-mcp  
**GitHub Stars:** 3  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/superops

## Description
MCP server for SuperOps unified PSA+RMM — clients, tickets, assets, and more.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "superops": {
    "command": "npx",
    "args": ["@wyre-ai/superops-mcp"],
    "env": {
      "SUPEROPS_API_TOKEN": "",
      "SUPEROPS_SUBDOMAIN": "",
      "SUPEROPS_REGION": ""
    }
  }
}
```

**Requires environment variables:** `SUPEROPS_API_TOKEN`, `SUPEROPS_SUBDOMAIN`, `SUPEROPS_REGION` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What the SuperOps MCP server does

The SuperOps MCP server gives MCP clients access to SuperOps.ai, a PSA/RMM platform. Its tools cover client records, service tickets, managed assets or endpoints, software and patch inventory, and technician information. Ticket tools support creating tickets, updating status or assignment, adding notes, and logging time.

The server also provides navigation and connectivity tools. Agents can select a domain, return to the main menu, or test access to the SuperOps API before making operational requests.

## How it works

The SuperOps MCP server communicates with SuperOps.ai through the platform’s GraphQL API. Its decision-tree design starts with domain navigation, while domain modules load when needed. This lazy-loading approach keeps the initial startup smaller and exposes tools relevant to the selected area.

Standard list operations use page-based pagination. Requests accept a 1-indexed `page` and a `pageSize` from 1 to 100, with 50 as the default. Responses include page information, total results, and a `hasMore` value. A `hasMore` value of `null` means there are no additional pages; it never returns `false`.

For operations that need API behavior not represented by the built-in tools, custom GraphQL query and mutation tools are available. Ticket retrieval can also render an interactive card in MCP Apps hosts such as Claude Desktop or Claude web. Its add-note action always creates an internal note rather than a public reply.

## Setup and configuration

Install the package from GitHub Packages with npm:

```bash
export NODE_AUTH_TOKEN=$(gh auth token)
npm install @wyre-ai/superops-mcp
```

GitHub Packages requires authentication even for this public package. Cloud deployments need a package token during the build: Cloudflare Workers uses `NODE_AUTH_TOKEN`, while DigitalOcean App Platform uses a build-time `GITHUB_TOKEN` secret.

The SuperOps MCP server requires these runtime variables:

- `SUPEROPS_API_TOKEN`: an API token generated from the API token tab in the SuperOps.ai profile settings.
- `SUPEROPS_SUBDOMAIN`: the tenant subdomain.
- `SUPEROPS_REGION`: `us` or `eu`.

A Claude Desktop configuration can launch the server with `npx` and pass these values in its `env` object.

## Tools and capabilities

Available tool groups include:

- Navigation and API connectivity testing.
- Client listing, lookup, and name search.
- Ticket listing, retrieval, creation, updates, notes, and time entries.
- Asset listing and retrieval, plus software inventory and patch status.
- Technician listing, detail lookup, and group listing.
- Custom GraphQL queries and mutations.

Ticket cards can be branded without rebuilding by using the documented `MCP_BRAND_*` variables or a `window.__BRAND__` injection. Other hosts receive the ordinary JSON behavior.

## Limitations and notes

The SuperOps API allows 800 requests per minute per API token. Filters use condition clauses with operators such as `is`, `contains`, `startsWith`, `includes`, and their negative forms. `equals` and `in` are not accepted. Filtering with an invalid field value, or against a JSON column instead of a nested path, can silently return zero rows.

For complex filter behavior, use the custom query tool. The repository includes a GraphQL schema and validates its documents against that schema during tests; the schema should be regenerated after an API change. The project is licensed under Apache-2.0.

_Full upstream README: https://allmcps.com/mcp/superops/readme_