> ## Documentation Index
> Fetch the complete documentation index at: https://learn.nexudus.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Telemetry

> What anonymous usage data the Nexudus MCP server collects, why, and how to ask for it to be turned off.

# Telemetry

The Nexudus MCP server records anonymous usage data so we can see which operations are useful, where things fail, and which AI assistants are connecting. Telemetry is **on by default** for the hosted server.

## What is collected

Each event captures only the **shape** of the call — never the arguments you sent or the data Nexudus returned.

| Field | Example | Description |
| - | - | - |
| `command` | `nexudus_list` | The MCP tool that was called. |
| `entity` | `bookings` | The Nexudus entity, when the call had one. |
| `operation` | `list` | The kind of operation (list, get, create, update, delete, run-command). |
| `success` | `true` | Whether the call succeeded. |
| `durationMs` | `245` | How long the call took in milliseconds. |
| `errorCategory` | `auth` / `not_found` / `validation` | A coarse classification of the error, when one occurred. |
| `caller` | `claude-ai/1.0.0` | The MCP client name and version reported during connection. |
| `traceId` | `abc123…` | A per-session identifier used to correlate calls in one session. |
| `version` | `5.0.0` | The MCP server version. |
| `os` | `linux` | The operating system of the server. |
| `isAgent` | `true` | Always `true` — every MCP caller is an AI agent. |
| `type` | `nexudus-mcp` | Distinguishes MCP events from CLI events in our analytics. |

## What is **not** collected

* **No tool arguments.** Field values you send (booking dates, member names, product prices, filter values) are never logged.
* **No response payloads.** The records returned by Nexudus are never logged.
* **No personal data about you.** Your name, email, and Nexudus account ID are not in the event.
* **No business data.** Names of your locations, members, or products are never sent.

## Where data is sent

When telemetry is enabled, events are shipped to **Logz.io**, a third-party analytics platform, over HTTPS with a 3-second timeout. If the network call fails for any reason — including Logz.io being unreachable — the event is silently dropped. Telemetry never adds latency to your tool calls and never causes a tool call to fail.

## Why we collect this

| Question we answer | Why it matters |
| - | - |
| Which entities and operations are used most? | Decide what to optimise or document better. |
| Where do AI assistants pick the wrong tool? | Improve the per-entity guides and tool descriptions. |
| Which MCP clients are most active? | Prioritise testing on those clients. |
| What is the typical response time? | Catch regressions in the server. |
| Which errors are most common? | Identify field-validation gaps or auth issues. |

## Turning telemetry off

Telemetry is configured at the server level, not per-user. If you are using the public hosted server at `https://mcp.nexudus.com` and want telemetry disabled for your traffic, contact Nexudus support — there is no in-client setting for it.

## Privacy summary

* **No sensitive data** — only the shape of calls is captured, never arguments or response data.
* **Failure-safe** — telemetry errors never affect tool calls.
* **Encrypted transmission** — all data is sent over HTTPS.
* **Aggregated reporting** — events are used to compute counts and trends, not to track individual users.

## Questions

If you have privacy concerns or want to opt out of telemetry on the hosted server, reach out to the Nexudus team or open an issue on the [project repository](https://github.com/Nexudus-Ltd/nexudus-coworking-mcp-stateful).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.