Kong Konnect MCP Server – API Gateway Management
Kong Konnect MCP Server is Kong's hosted Model Context Protocol server for interacting with Kong Konnect from MCP-compatible IDEs, assistants, and command line clients. It lets agents discover Konnect API operations, inspect schemas, execute authorized actions, query gateway analytics, debug performance issues, and search Kong documentation while respecting Konnect permissions.
Overview
Kong Konnect MCP Server is the official hosted MCP interface for Kong Konnect, Kong's SaaS API management platform. It gives MCP-compatible AI assistants and IDE copilots a structured way to work with Konnect resources from the developer environment instead of switching between chat, documentation, dashboards, and API tooling.
The server is designed for API infrastructure workflows. It connects to Konnect backend services to retrieve gateway configuration, analytics data, active tracing information, and Kong documentation. The same tool layer is also used by KAi, Kong's in-product AI assistant for Konnect.
What the MCP server enables
Kong documents the current hosted MCP server around a discover-then-execute model rather than a large fixed set of individual hardcoded tools. The core tools are search, get_schema, and execute: an assistant can search for the relevant Konnect API operation, fetch the operation schema it needs, and then execute the operation with the correct parameters.
Through that workflow, agents can support Konnect entity management for control planes, services, routes, consumers, consumer groups, plugins, and vaults. They can also query API request analytics with filters such as time range, status codes, consumers, services, and routes; create debug sessions with active tracing to investigate performance issues; and search Kong documentation for configuration or troubleshooting guidance.
When to use it
Use this MCP server when a developer, platform engineer, or API operator wants an AI assistant to help manage or inspect Kong Konnect resources from an IDE, terminal, or MCP-aware assistant. Practical examples include asking an assistant to find services in a control plane, investigate 5xx traffic for a route, identify consumers affected by an issue, inspect plugin configuration, or look up relevant Kong documentation while troubleshooting.
Connection and authentication
Kong Konnect MCP Server is a remote HTTP MCP service. Kong provides regional endpoints for the United States, Europe, Australia, and India, and the client should use the endpoint for the region where the Konnect resources are deployed. Authentication uses a bearer token in the Authorization header. Kong supports Personal Access Tokens for user-scoped access and System Account Access Tokens for automated workflows. Access follows the permissions, regional boundaries, and role-based access controls associated with the token.
Key considerations
Kong states that Konnect MCP Server is incompatible with on-prem deployments. MCP access is enabled by default for organizations, but administrators can disable it in organization settings. Organizations using multiple Konnect regions need separate MCP server connections for each region because resources cannot be accessed across regions from a single connection. Tool failures should be handled as permission, parameter, regional endpoint, rate-limit, or missing-resource issues rather than assumed model errors. Store tokens in the MCP client's secure secret mechanism whenever available and avoid hardcoding Personal Access Tokens in shared configuration files.
Supported Transports
streamable_http
URL: https://us.mcp.konghq.com/
streamable_http
URL: https://eu.mcp.konghq.com/
streamable_http
URL: https://au.mcp.konghq.com/
streamable_http
URL: https://in.mcp.konghq.com/
Frequently Asked Questions
- When should an AI agent use Kong Konnect MCP Server?
- Use it when an agent needs to help manage or inspect Kong Konnect API infrastructure from an MCP-compatible client, such as an IDE, Claude Code CLI, GitHub Copilot, Cursor, or another assistant. Kong documents use cases including Konnect resource management, API performance debugging, analytics and monitoring, and Kong documentation search.
- What does Kong Konnect MCP Server add to an AI agent's capabilities?
- It gives the agent live, permission-aware access to Kong Konnect operations and context. The agent can search for relevant Konnect API operations, retrieve operation schemas, execute authorized requests, query gateway analytics, work through active tracing debug sessions, and search Kong documentation instead of relying only on static model knowledge.
- What can the AI agent access or manage through this server?
- Kong documents support for Konnect entity management across resources such as control planes, services, routes, consumers, consumer groups, plugins, and vaults. It can also query API request data with filters for time range, status codes, consumers, services, and routes. Actual access depends on the PAT or System Account Access Token permissions and the selected Konnect region.
- How is authentication configured for Kong Konnect MCP Server?
- Authentication is configured with an HTTP bearer token in the Authorization header. Kong supports a Personal Access Token for user-scoped access or a System Account Access Token for automated workflows. The token should be stored in a secure client secret prompt or environment variable such as KONNECT_ACCESS_TOKEN rather than embedded directly in shared files.
- Which transport should be used for Kong Konnect MCP Server?
- Use Kong's hosted remote HTTP MCP endpoints. Choose the regional endpoint that matches where the Konnect resources are deployed: US, EU, AU, or IN. Kong's older local GitHub repository for mcp-konnect is deprecated, so this YAML lists only the current hosted HTTP MCP endpoints documented by Kong.