The Freshworks Model Context Protocol (MCP) server enables secure integration between AI tools and Freshdesk. It acts as a middleware layer that exposes selected Freshdesk capabilities, such as ticket management, contacts, and knowledge base operations, through authenticated API interactions.

The MCP server relies entirely on Freshdesk’s existing authentication and authorization systems. It does not introduce a separate identity model and does not grant access beyond what the authenticated user or API key is permitted to perform.

Info: For an overview of MCP integration, including setup and configuration steps, see the MCP integration article. 

TABLE OF CONTENTS

• Deployment and isolation
• Authentication
• Authorization and permissions
• Tool exposure and controls
• Data handling
• Logging and audit
• Input validation
• Network security
• Usage limits and safeguards
• Security considerations for AI integrations

Deployment and isolation

The Freshworks MCP server is exposed as a tenant-specific endpoint:

https://<your-freshdesk-domain>/mcp

Each Freshdesk account operates within its own isolated environment.

• All requests are scoped to the tenant’s Freshdesk instance.

• Data and tool execution are isolated per tenant.

• The MCP server functions as a proxy layer over Freshdesk APIs and does not maintain independent state.

Authentication

Authentication is required for every MCP request.

Freshworks MCP for Freshdesk currently supports API key authentication only.

Each request must include a valid API key in the authorization header. Requests without valid credentials are rejected.

Note: OAuth authentication is not currently supported for Freshdesk MCP integrations.

Authorization and permissions

The MCP server does not implement its own authorization layer. All access control is enforced using Freshdesk’s built-in role-based permissions.

• Only users with administrator or agent roles can access MCP capabilities.

• Actions performed through MCP inherit the permissions associated with the API key.

• Users can only perform operations already permitted via Freshdesk APIs.

Examples:

• A user without ticket update permissions cannot modify tickets via MCP.

• API keys scoped to limited roles restrict available tool operations accordingly.

Tool exposure and controls

The MCP server exposes a defined set of tools aligned with Freshdesk functionality, including:

• Ticket and conversation management

• Contacts, agents, and groups

• Company management

• Knowledge base (solutions) operations

Tool usage is governed by:

• Plan-based limits (rate limits and monthly quotas)

• Permission scopes derived from API key access

• Client-side controls in AI tools

During the Early Access Program (EAP), all tools are available under controlled limits.

Data handling

The MCP server processes data dynamically via Freshdesk APIs.

• No independent data storage or caching is performed.

• Data returned is limited to what the authenticated identity can access.

• All operations (read and write) are executed directly against Freshdesk services.

This ensures:

• Data consistency with the Freshdesk UI and APIs

• Minimal data exposure

• No duplication of sensitive information

Logging and audit

MCP interactions are captured within Freshdesk’s existing logging and audit systems.

• Actions performed through MCP are recorded similarly to API and UI operations.

• Changes (for example, ticket updates or contact creation) appear in activity history.

• Errors and failed authentication attempts can be traced for troubleshooting.

Sensitive data handling follows Freshdesk’s logging policies.

Input validation

All inputs to MCP tools are validated before execution.

• Malformed or invalid requests are rejected.

• Structured error responses are returned to the client.

• Requests must comply with Freshdesk API schemas and constraints.

This prevents unintended operations and improves system reliability.

Network security

All communication with the Freshdesk MCP server is secured over HTTPS.

• Data is encrypted in transit using TLS.

• Requests must include valid authentication headers.

• Communication occurs within Freshdesk’s secure infrastructure.

Usage limits and safeguards

Freshdesk MCP integration is currently available under an Early Access Program (EAP).

• Available only on the Enterprise plan

• Subject to rate limits and monthly usage caps

• Limits may change upon general availability

These safeguards help maintain platform stability and prevent misuse.

Security considerations for AI integrations

When integrating AI tools with Freshdesk MCP:

• Store API keys securely and avoid exposing them in client-side code.

• Rotate API keys periodically.

• Restrict access to trusted users and systems.

• Monitor usage and audit logs regularly.

• Apply least-privilege principles when assigning roles.