Introduction
One of the most powerful features of Introw’s AI Agent is its ability to connect with external knowledge bases and tools via an MCP (Model Context Protocol) server.
MCP is an open standard that allows AI agents to securely access and query data from external sources. By linking your MCP server to Introw, you can extend the AI Agent beyond Introw’s built-in resources making it smarter, more relevant, and more aligned with your organization’s unique needs.
What is MCP?
The Model Context Protocol (MCP) is a universal framework that connects AI systems with external data, tools, and services. It standardizes how AI queries information and retrieves results, ensuring security, consistency, and compatibility across platforms.
In the context of Introw, MCP enables the AI Agent to:
Pull in external knowledge base content
Connect to custom enterprise tools or APIs
Provide more personalized, accurate answers to your partners and team
How MCP Works in Introw
Introw supports direct integration with external MCP servers through a dedicated MCP configuration.
Once configured:
The Introw AI Agent functions as an MCP client, issuing structured, authenticated requests to the configured MCP server.
Requests are sent over TLS-encrypted channels and include the bearer token for access control.
The MCP server executes the request, retrieves or computes the required data, and returns a structured MCP response.
The AI Agent consumes the MCP response and incorporates the returned data into its reasoning and final response delivered to the partner or end user.
All MCP traffic is handled in real time, with end-to-end encryption in transit and credentials securely stored and managed via the Introw UI.
Connecting Your MCP Server to Introw
Follow these steps to connect your MCP server to Introw:
Prepare your MCP Server
Ensure your knowledge base or data source is exposed via an MCP-compatible endpoint.
Configure authentication using a bearer token or API key.
Add the MCP Connection
Specify the MCP server URL and authentication credentials (bearer token).
The connection configuration is securely stored and used by the AI Agent at runtime.
Validate the Integration
Use the Test Agent to issue sample queries against the MCP server.
Confirm that responses are returned successfully and contain the expected data.
Go Live
After validation, the AI Agent begins using the MCP server for partner-facing and internal queries.
Security & Privacy
Security is central to Introw’s MCP integration. Key protections include:
Authentication: The AI Agent uses secure API tokens or keys to communicate with your MCP server. Unauthorized clients cannot connect.
Encryption: All queries and responses travel over encrypted HTTPS/TLS channels, preventing interception or tampering.
Data segregation: Your organization’s MCP data is isolated from all other tenants. No other company or partner can access your server.
Additionally, Introw enforces prompt protection and malicious input filtering to ensure attackers cannot manipulate the AI Agent into exposing or misusing MCP-connected data.
Benefits of connecting your MCP
By integrating your MCP server, you can:
Extend knowledge: Incorporate external data sources into the AI Agent’s responses
Personalize support: Surface company-specific information for partners
Unify access: Bring multiple knowledge systems into one conversational interface
Stay secure: Leverage enterprise-grade encryption, isolation, and compliance safeguards