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

# Privacy and Security

> See where your requests go, what stays on your Mac, and how to control context, credentials, tools, analytics, and diagnostics.

Enconvo stores your data on your Mac and uses online services for the features and providers you choose. Where a request goes depends on the feature, the provider you pick, the tools you turn on, and the integrations you connect. This page shows each flow and the controls for it.

## Where your content goes

| Action | What is sent, and where |
| - | - |
| Cloud AI chat | Your prompt, the conversation context, and the attachments you add, to the model provider |
| Enconvo Cloud content check, from November 14, 2026 | The text you just wrote, to OpenAI's moderation model, before a Cloud model gets it |
| Cloud speech recognition | Captured audio, to the speech service you picked |
| Cloud text-to-speech | Text, to the voice service you picked |
| Cloud embeddings or reranking | The document and query content needed for retrieval |
| Web search or fetch | The search query or URL, to the search service or website |
| Local model inference | Stays on your Mac: the model runs there once its files are downloaded |
| Browser or computer tools | Whatever content and actions the tool's granted access allows |
| IM channel | Messages between the connected messaging service and the Agent |
| Sync | Your chats, settings, and knowledge bases, to your Enconvo account or your iCloud, once you turn Sync on. API keys leave your Mac only encrypted, once you turn them on. |

A local chat model alone doesn't make every tool offline. A web search, a remote MCP tool, a cloud transcription provider, or a cloud embedding model still sends content over the network.

## Enconvo Cloud's usage rules

From November 14, 2026, Enconvo Cloud checks each message against our [Terms of Use](https://www.enconvo.com/terms) before it reaches a Cloud model. Only the text you just wrote is checked, not the earlier conversation, your files, or your images. Nothing changes when you use your own API keys or local models.

* **What's kept:** If a message is refused or flagged, Enconvo keeps the account, time, model, and scores for up to 90 days, and never the words.
* **A refused message:** It doesn't reach the model and costs no points. Your chat shows why.
* **Repeated refusals:** They can pause Cloud AI on your account. You can still sign in, manage or cancel your subscription, and delete your account. A subscription keeps running until you cancel it.
* **If we got it wrong:** Write to [support@enconvo.com](mailto:support@enconvo.com).

So that a provider's safety action applies to one account instead of everyone, Enconvo Cloud also sends OpenAI and Anthropic a one-way identifier made from your account with each request. It isn't your email address. The [Privacy Policy](https://www.enconvo.com/privacy) has the details.

## Control what you send

Look over attachments and context before you send a request. Appshots, screenshots, selected text, clipboard content, and browser pages can hold more than the sentence you mean to ask about.

Choose your context sources in [Context Awareness](https://docs.enconvo.ai/features/context-awareness.md). Grant [macOS permissions](https://docs.enconvo.ai/configuration/permissions.md) only for the features you use, and revoke them in System Settings when you no longer need them.

## Credentials

You manage API keys and account sign-ins on each provider's **Authentication** section and in [Credential Management](https://docs.enconvo.ai/configuration/credentials.md). Treat your local Enconvo configuration, backups, exported settings, and support logs as sensitive, and keep keys and account tokens out of prompts, screenshots, and shared files.

A provider profile can share its credentials across several features. Check every feature that uses a profile before you delete or replace it.

> [!WARNING]
>
> Removing a key from Enconvo doesn't revoke it. If a key leaks, revoke it on the provider's own site.

## Tools and integrations

An Agent can act through the tools you turn on. Browser control, computer use, shell commands, file edits, Gmail, and external MCP servers each reach different things. Check the Agent's tools, and the action it proposes, when it works with important data.

Running an extension in its own worker doesn't stop it from reaching your local files or the network. Install tools from sources you trust and check their configuration.

## Local AI clients

[Knowledge Base MCP](https://docs.enconvo.ai/advanced/knowledge-base-mcp.md) is off until you turn it on, and accepts local connections. Its visibility switches choose which knowledge bases it exposes. A local client may send what it retrieves to its own AI provider, so set up that client with care too.

## Analytics and diagnostics

Enconvo records product usage events under a device ID, not your account. The events are designed to leave out email addresses, conversation text, and media. To stop sharing them, turn off **Share anonymous product analytics** under **Privacy** in **Settings → General**. Crash and freeze reports are sent only while that switch is on, and carry none of your content.

Diagnostic logs are separate from analytics. When Enconvo freezes, crashes, or misbehaves, send them to the Enconvo team with **Submit Diagnostics…** in the menu bar menu, or from **Settings → Diagnostics**:

* Before anything leaves your Mac, Enconvo takes out personal details, such as your home folder's name, file names, email addresses, API keys, tokens, links, IP addresses, and the text of prompts and replies.
* A report carries a random ID made only for diagnostics, not your account or email. Only the Enconvo team can read it, and it's deleted after a set number of days.
* Logs can still name the apps, commands, and models you used.
* The page shows exactly what was sent, and lets you withdraw or delete each report.

When you share a log or screenshot some other way, look it over first and remove private prompts, paths, keys, account details, and document contents that aren't needed.

## Keep or remove local data

Each feature manages its own data: chat history, knowledge bases, recordings, and credentials each have their own controls. Read deletion confirmations, and keep backups of anything you need.

Removing the app doesn't always remove its configuration and data. Removing local data doesn't delete copies an external service already processed. For a provider's retention and account policies, see that provider: this page covers Enconvo's data flows, not theirs.

## Related

> [card] [macOS Permissions](https://docs.enconvo.ai/configuration/permissions.md)
>
> What each permission lets Enconvo do.

> [card] [Credentials](https://docs.enconvo.ai/configuration/credentials.md)
>
> API keys, sign-ins, and shared profiles.

> [card] [Context Awareness](https://docs.enconvo.ai/features/context-awareness.md)
>
> Choose what context goes with a request.

> [card] [Enconvo Account](https://docs.enconvo.ai/configuration/account.md)
>
> Sync, and what it keeps on your computers.


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