---
title: "Enhanced Security and Control: Unified MCP Server Gets Security Upgrade"
img: https://s3.us-east-2.amazonaws.com/unified-article-images/enhanced_security_and_control_unified_mcp_server_gets_security_upgrade-icon.png
date: 2025-07-15T00:00:00.000Z
tag: Product
description: "We're excited to announce a significant upgrade to the Unified MCP (Model Context Protocol) server that brings enhanced security controls and granular..."
url: "https://unified.to/blog/enhanced_security_and_control_unified_mcp_server_gets_security_upgrade"
---

# Enhanced Security and Control: Unified MCP Server Gets Security Upgrade
------
_July 15, 2025_

# Enhanced Security and Control: Unified MCP Server Gets Security Upgrade


We're excited to announce a significant upgrade to the [Unified [MCP](/mcp)](/mcp) ([Model Context Protocol](/blog/what_is_model_context_protocol_mcp_a_guide_for_product_managers)) server that brings enhanced security controls and granular permission management to your AI Agent integrations. These new features give you unprecedented control over data access and privacy when connecting any Unified connection to LLM providers.


## What's New: Advanced Permission Controls


### Granular Data Access Management


The upgraded MCP server introduces sophisticated permission controls that operate at two levels:


**Connection Categories and Permissions**: Each integration connection has permissions or categories set by our customers to control what access their use-case requires. This configuration is utilized to determine which tools are available in the MCP server. 


**URL Parameter Permissions**: For even finer control, you can now use the new `permissions` URL parameter to restrict access at the API level. This is particularly powerful for implementing read-only access patterns.  The values for this new parameter are the same as the unified scope permissions used to authorizing connections.


Using the new `permissions` parameter, you can restrict your MCP server to only read existing data from the connection, preventing any write, create, update, or delete operations.


**Example:**


```plain text
https://mcp-api.unified.to/mcp?token={token}&permissions=hris_employee_read
```


This ensures that your AI agents can analyze and retrieve information without the ability to modify or create new records, providing an essential safety layer for sensitive environments.


### Enhanced Data Privacy with Sensitive Data Hiding


Privacy protection gets a major boost with the new `hide_sensitive` parameter. This feature automatically filters out sensitive information from API responses, ensuring that confidential data never reaches your LLM providers.


**Example Usage:**


```plain text
https://mcp-api.unified.to/mcp?token={token}&hide_sensitive=true
```


When enabled, the MCP server will:

- Automatically detect and remove personally identifiable information (PII)
- Filter out financial data, social security numbers, and other sensitive fields
- Can ensure compliance with data protection regulations

## What's Also New: Structured Data in MCP v2025-06-18


Along with these new security and privacy enhancements, we've also upgraded the Unified MCP server to support the latest MCP server version (v2025-06-18), which now supports returning real 'structured' data and not just text responses.


Previously, MCP servers were limited to returning plain text responses, which required LLMs to parse and interpret unstructured data. With MCP v2025-06-18, the Unified MCP server can now return properly structured data objects, including:

- **JSON objects** with nested properties and arrays
- **Typed data fields** that preserve data types (numbers, booleans, dates)
- **Relational data structures** that maintain relationships between entities

### Enhanced AI Understanding and Performance


This structured data capability dramatically improves how AI models work with your integrated data:


**Improved Accuracy**: LLMs can now directly access structured data fields without parsing text, reducing interpretation errors and improving response accuracy.


**Better Performance**: Structured responses are processed faster by AI models, leading to quicker response times and more efficient API usage.


**Richer Interactions**: AI agents can now perform more sophisticated operations on your data, such as sorting, filtering, and aggregating results with greater precision.


**Enhanced Context**: The structured format provides LLMs with better understanding of data relationships and field meanings, leading to more intelligent responses.


## What's Next


This upgrade represents our commitment to making data integrations for AI agents both powerful and secure. We will continue to enhance the Unified MCP server with additional security features, performance improvements, and expanded integration support.


The Model Context Protocol is evolving rapidly, and we're staying at the forefront of these developments to ensure our customers have access to the latest capabilities while maintaining the highest security standards.


---


**Ready?** The enhanced Unified MCP server is available now at the same endpoints. Simply add the new permission parameters to start taking advantage of these powerful new security features.


For technical documentation and implementation details, visit our [MCP Server documentation](https://docs.unified.to/mcp).


_Questions about implementing these new features?_ [_Contact our team_](http://unified.to/contact) _for personalized guidance._