Follow these steps to run the Kong MCP server and tools inside a Meshagent Room for temporary, cloud-based testing:

​
Required Environment Variables:

  • KONNECT_REGION (e.g., us) – The Konnect region.
  • KONNECT_ACCESS_TOKEN (e.g., kpat_api_key_here) – Your Konnect API key.

​
Steps

  1. Install Meshagent CLI with all extras: 
    pip install "meshagent[all]"
  2. Sign up and authenticate Meshagent:
    Instructions: Meshagent CLI Getting Started
  3. Start the MCP server in a test Room: 
    meshagent service test --room=test --role=agent --image=meshagent/mcp-kong:latest \
  --env MESHAGENT_PORT=8001 \
  --env KONNECT_REGION=us \
  --env KONNECT_ACCESS_TOKEN=kpat_api_key_here \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-kong-service-test
    • Creates a Meshagent Room with a test MCP server callable via Room tools. Room is auto-destroyed if inactive.
  4. Start the chatbot (tool-enabled) in the Room: 
    meshagent chatbot join --room=test --agent-name=mcp-kong --name=mcp-kong --toolkit=mcp-kong
    • Bot runs in the Room, using the Kong toolkit.
    • Multiple toolkits can be attached to the same Room & agent.
    • Room access link is provided on success.
  5. Try it out:
    Open the provided Room link in your browser and chat with the agent to run queries or actions on your MCP server tools.

​
Project Level Deployment (Production Usage)

For persistent, always-on MCP tools and chatbot agents in every Room within a project (recommended for production):
  1. Create persistent agent service at project level: 
    meshagent service create --role=agent --image=meshagent/mcp-kong:latest \
  --env MESHAGENT_PORT=8001 \
  --env KONNECT_REGION=us \
  --env KONNECT_ACCESS_TOKEN=kpat_api_key_here \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-kong-service
  2. Create persistent chatbot service at project level: 
    meshagent service create --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-kong-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-kong-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-kong-chatbot --toolkit=mcp-kong"
  • These services will auto-join every Room created inside the Meshagent project, making the MCP tools and chatbots available by default—no local/server process needed.

​
Tools Available

A Model Context Protocol (MCP) server for interacting with Kong Konnect APIs, allowing AI assistants to query and analyze Kong Gateway configurations, traffic, and analytics.

​
Tools Provided

ToolShort Description
check_control_plane_group_membershipCheck if a control plane is a member of any group.
get_consumer_requestsRetrieve and analyze API requests made by a specific consumer.
get_control_planeGet detailed information about a specific control plane.
list_consumersList all consumers associated with a control plane.
list_control_plane_group_membershipsList all control planes that are members of a specific control plane group.
list_control_planesList all control planes in your organization.
list_pluginsList all plugins associated with a control plane.
list_routesList all routes associated with a control plane.
list_servicesList all services associated with a control plane.
query_api_requestsQuery and analyze Kong API Gateway requests with customizable filters.

​
Tool Details

​
check_control_plane_group_membership

Check if a control plane is a member of any group.
  • Inputs: controlPlaneId (string) – ID to check membership for.
  • Outputs: Membership status, group info, conflicts, related tools.

​
get_consumer_requests

Retrieve and analyze API requests from a specific consumer.
  • Inputs: consumerId (string), timeRange (string), successOnly, failureOnly, maxResults.
  • Outputs: Metadata, usage statistics (latency, success, etc.), detailed request list.

​
get_control_plane

Get detailed information about a specific control plane.
  • Inputs: controlPlaneId (string)
  • Outputs: All control plane metadata, related endpoints, labels, cluster type, and more.

​
list_consumers

List all consumers associated with a control plane.
  • Inputs: controlPlaneId (string), size, offset
  • Outputs: Consumer metadata, paginated consumer list, related tools.

​
list_control_plane_group_memberships

List all control planes that are members of a group.
  • Inputs: groupId (string), pageSize, pageAfter
  • Outputs: Member control plane metadata, membership status, conflicts.

​
list_control_planes

List all control planes in your organization.
  • Inputs: pageSize, pageNumber, filterName, filterClusterType, filterCloudGateway, labels, sort
  • Outputs: Array of control planes, filters, sort info, overall usage hints.

​
list_plugins

List all plugins associated with a control plane.
  • Inputs: controlPlaneId (string), size, offset
  • Outputs: Plugin details, types, config, protocols, scoping, tags.

​
list_routes

List all routes associated with a control plane.
  • Inputs: controlPlaneId (string), size, offset
  • Outputs: Route metadata, HTTP methods, paths, hosts, service links, flags.

​
list_services

List all services associated with a control plane.
  • Inputs: controlPlaneId (string), size, offset
  • Outputs: Service ID, endpoint, protocol, retries, timeouts, tags.

​
query_api_requests

Query and analyze Kong API Gateway requests.
  • Inputs: timeRange, filters (statusCodes, excludeStatusCodes, httpMethods, consumerIds, serviceIds, routeIds), maxResults.
  • Outputs: Filtered request metadata and full detail per request.

​
Additional Resources