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

# Security and data handling

> How Condense limits what an AI tool can do, what leaves Condense when you connect one, and how to run AI tools safely.

Connecting an AI tool to Condense does not give anyone new access. It does bring two things you should know about before you start.

* **Data leaves Condense.** What a tool returns, such as a pipeline, a configuration or a log, goes to your AI tool, and from there to the model provider that tool uses. Use AI tools that your organization has approved.
* **One tool changes your environment.** `deploy_connector` can deploy a connector, so an AI tool that misunderstands a request can deploy something you did not intend.

## How Condense limits access

| Control | What it does |
| :- | :- |
| **Sign-in with your Condense account** | You connect through OAuth in your browser. The AI tool works as you. |
| **Your Condense role** | The AI tool inherits your permissions. It cannot do what you cannot do. |
| **One environment per connection** | You choose the environment on the approval page. |
| **Tools you approve** | You choose which tools the AI tool may use, and you can change that later on the AI Tools page. |
| **Operating mode** | An Organization Admin can set an environment to Strict read-only, which blocks any action that modifies state. |
| **Allowed tools** | An Organization Admin chooses which tool groups and tools AI tools may use in each environment. |
| **Pause, resume and disconnect** | You can end your own connections. Administrators can pause, resume or disconnect any AI tool or member. |
| **Usage limits** | Usage is limited per organization, with a 5-hour limit and a daily limit. |
| **Terms and Privacy Policy** | You agree to both on the approval page before you accept. |

## Risks and how to reduce them

### Text your AI tool reads can try to steer it

Logs, configurations and documentation are text. A line in a log could contain words written to push an AI tool into doing something you did not ask for. Treat what an AI tool reads from your environment as data, not as instructions, and be careful about what it does next, especially when it can deploy.

### Recommendations

**1. Start with a non-production environment**

Connect to a development or test environment first. Choose the environment on the approval page.

**2. Use Strict read-only where AI tools should only look**

Administrators can set an environment to Strict read-only in [AI Governance](https://docs.condense.io/condense/v3.0.0/mcp/ai-governance.md).

**3. Approve only what the AI tool needs**

Clear the tools it does not need. If it only reads, clear `deploy_connector`.

**4. Check what it is about to do**

Read the tool calls your AI tool asks you to approve, especially a deployment, and read the configuration before you confirm it.

**5. Keep secrets out of the chat**

Connector configurations can include broker addresses, usernames and passwords. Keep real secrets out of your prompts where you can.

**6. Disconnect what you do not use**

Remove AI tools you no longer use from the [AI Tools page](https://docs.condense.io/condense/v3.0.0/mcp/your-connected-ai-tools.md).

### For administrators

* Review [AI Access Management](https://docs.condense.io/condense/v3.0.0/mcp/ai-access-management.md) regularly to see which AI tools members have connected and how much they are used.
* If you suspect a problem, pause first. A pause keeps the connection so you can resume it, and a disconnect makes members connect again.
* Set the operating mode and the allowed tools for each environment in [AI Governance](https://docs.condense.io/condense/v3.0.0/mcp/ai-governance.md) before members connect.


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