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

# Privacy

> Account-wide controls over where your prompts go and what is recorded.

nemonist forwards each request through a model router to the model's provider, which answers it; the router is named in the [privacy policy](https://nemonist.com/privacy). nemonist keeps your workspace's memory notes and the requests that made them, and erasing the project erases both (see [Memory](https://nemonist.com/docs/concepts/memory.md#forget-and-erase)).

The **Privacy** page applies to every request in the account. Keys can be restricted further with their own rules and [guardrails](https://nemonist.com/docs/concepts/guardrails.md).

* **No training or publishing.** Never route to endpoints that may train on or publish your prompts.
* **Disable Anthropic first-party.** Claude models are still served through Amazon Bedrock and Google Vertex.
* **Disable OpenAI first-party.** OpenAI models are still served through Azure.
* **Allowed and ignored providers.** Only allowed providers serve requests (empty allows every provider); ignored providers never do.
* **Don't record project folder names.** Projects on the [Memory](https://nemonist.com/docs/concepts/memory.md) page are then named by their files, or by a name you give them.
* **Maximum API key lifetime.** New keys must expire within this window, and requests from existing keys that never expire or expire later are refused.

nemonist's own memory is stored in your account either way and is never used for training. See the [privacy policy](https://nemonist.com/privacy).


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