RecordPoint MCP Server

  • Updated

The RecordPoint MCP Server enables secure, real-time connections between AI tools and your governed data in RecordPoint.

MCP (Model Context Protocol) is an open standard that allows AI systems to securely connect to external data sources. Instead of uploading documents into an AI model or creating static snapshots, MCP provides a live, permission-aware bridge between AI and your governed data platform.

RecordPoint’s MCP Server ensures that:

  • All data accessed by AI is already classified, tagged, and governed.

  • Access is enforced using Attribute-Based Access Control (ABAC).

  • Responses are permission-aware and user-specific.

  • Data remains live and up to date not a snapshot from last week.

This creates a two-way connection between your data sources and AI-powered tools, while maintaining governance, security, and compliance controls.

Purpose

The RecordPoint MCP Server acts as a secure bridge between AI systems and your RecordPoint Governed AI Pipeline (RGAP).

Rather than pointing a model at raw content and hoping for the best, the MCP Server ensures:

  • The AI only sees data the user is permitted to access.

  • Access is enforced in real time using ABAC rules.

  • Sensitive content remains protected.

  • All interactions are logged for audit and compliance purposes.

If two different users ask the same question, they may receive different answers — or no answer at all — depending on their permissions. This is least-privilege access enforced automatically, with no additional configuration required.

How It Works

Feature 1 - Connect Any MCP-Compatible AI

Connect MCP-compatible AI tools to your governed data instantly — without custom builds or complex integrations.

Supported AI tools include:

  • ChatGPT (Apps + Developer Mode required)

  • Claude (Custom Connectors supported plans)

  • Other MCP-compatible AI systems and custom agents

Feature 2 – Permission-Aware AI Access

  • Data is live and reflects the current state of your RecordPoint environment.

  • ABAC rules are enforced automatically.

  • The AI never receives access to data outside the user’s permissions.

  • No data snapshots or uncontrolled document uploads.

If another user with different permissions connects, they will receive different results — or no results — based on their access level.

Feature 3 – Control, Visibility & Compliance

All AI access through MCP is:

  • Permission-based

  • Role-aware

  • Fully logged

This ensures alignment with internal governance requirements and regulatory frameworks such as:

  • EU AI Act

  • NIST AI RMF

  • ISO standards and internal compliance policies

Pre-Requisites 

RecordPoint MCP requirements

To use the RecordPoint MCP Server:

  • You must have an active account in the RecordPoint Data Governance platform.

  • MCP Server is currently available in:

    • PAUE

    • PUSW

    • PUKS
      (Private GEOs and Canada coming soon.)

  • You must be:

    • A RecordPoint Admin, or

    • An AI Governance Admin + RecordPoint Visitor

These roles are required to access:

  • RecordPoint Governed AI Pipeline (RGAP)

  • RecordPoint MCP Server
     

RGAP (RecordPoint Governed AI Pipeline) 

To enable AI Governance (RGAP), the following capabilities must be turned on:

  • Semantic Search

  • Elastic Search

  • Binary Protection

  • Content Search

  • Enterprise Reporting & Analytics

Optional (recommended):

  • Content Signals (enhances project review reporting)

Important:
Content Sampling should NOT be enabled if you want to access RGAP, as it prevents destruction of binaries/cracked text after classification.

Setup - Using MCP

Connecting MCP to ChatGPT

Supported Plans:
ChatGPT plans that support Apps and Developer Mode (e.g., Team or Enterprise plans).

Steps:

  1. Open ChatGPT.

  2. Go to Settings.

  3. Navigate to Advanced Settings and turn on Developer Mode.

  4. Go to Apps.

  5. Click Create App.

  6. Enter the MCP URL from your RGAP Project.

  7. Authenticate using your RecordPoint credentials.

  8. Save and confirm the connection.

Once connected, ChatGPT will securely retrieve governed data in real time, based on the logged-in user’s permissions. All access remains ABAC-enforced and fully logged.

Connecting MCP to Claude

Supported Plans:
Claude plans that support Connectors and custom connector configuration (e.g., Team or Enterprise plans).

Steps:

  1. Open Claude.

  2. Go to Settings.

  3. Select Connectors.

  4. Click Add Custom Connector.

  5. Enter the MCP URL from your RGAP Project.

  6. Authenticate using your RecordPoint credentials.

  7. Save and confirm the connection.

Once connected, Claude will securely retrieve governed data in real time, based on the logged-in user’s permissions. All access remains ABAC-enforced and fully logged.

How MCP Differs from Document Uploads

Traditional approach:

  • Upload documents into an AI tool.

  • Create static data snapshots.

  • Manually manage access controls.

  • Risk exposing sensitive information.

With RecordPoint MCP:

  • Data remains in RecordPoint.

  • AI retrieves only what the user is permitted to see.

  • Access controls are enforced automatically.

  • Data is always live and up to date.

Notes

  • MCP access is user-specific and permission-aware.

  • Different users may receive different answers based on ABAC rules.

  • Data is not exported or permanently copied into the AI system.

  • All access is logged for security and compliance reporting.

  • Private GEO and Canada availability is coming soon.

     

Need Help?

For assistance with RGAP setup and configuration:

Was this article helpful?

0 out of 0 found this helpful

Have more questions? Submit a request