feat: add troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota (#42)

Author: tembleking

AGENTS.md

@@ -60,6 +60,7 @@ The handler filters tools dynamically based on `GetMyPermissions` from Sysdig Se
 | `troubleshoot_kubernetes_list_top_network_errors_in_pods` | `tool_troubleshoot_kubernetes_list_top_network_errors_in_pods.go` | Shows the top network errors by pod over a given interval. | `promql.exec` | "Show the top 10 pods with the most network errors in cluster 'production'" |
 | `troubleshoot_kubernetes_list_count_pods_per_cluster` | `tool_troubleshoot_kubernetes_list_count_pods_per_cluster.go` | List the count of running Kubernetes Pods grouped by cluster and namespace. | `promql.exec` | "List the count of running Kubernetes Pods in cluster 'production'" |
 | `troubleshoot_kubernetes_list_underutilized_pods_by_cpu_quota` | `tool_troubleshoot_kubernetes_list_underutilized_pods_by_cpu_quota.go` | List Kubernetes pods with CPU usage below 25% of the quota limit. | `promql.exec` | "Show the top 10 underutilized pods by CPU quota in cluster 'production'" |
+| `troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota` | `tool_troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota.go` | List Kubernetes pods with memory usage below 25% of the limit. | `promql.exec` | "Show the top 10 underutilized pods by memory quota in cluster 'production'" |
 
 Every tool has a companion `_test.go` file that exercises request validation, permission metadata, and Sysdig client calls through mocks.
 Note that if you add more tools you need to also update this file to reflect that.

README.md

@@ -163,6 +163,11 @@ The server dynamically filters the available tools based on the permissions asso
   - **Required Permission**: `promql.exec`
   - **Sample Prompt**: "Show the top 10 underutilized pods by CPU quota in cluster 'production'"
 
+- **`troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota`**
+  - **Description**: List Kubernetes pods with memory usage below 25% of the limit.
+  - **Required Permission**: `promql.exec`
+  - **Sample Prompt**: "Show the top 10 underutilized pods by memory quota in cluster 'production'"
+
 ## Requirements
 
 - [Go](https://go.dev/doc/install) 1.25 or higher (if running without Docker).

cmd/server/main.go

@@ -106,6 +106,7 @@ func setupHandler(sysdigClient sysdig.ExtendedClientWithResponsesInterface) *mcp
 		tools.NewTroubleshootKubernetesListTopNetworkErrorsInPods(sysdigClient),
 		tools.NewTroubleshootKubernetesListCountPodsPerCluster(sysdigClient),
 		tools.NewTroubleshootKubernetesListUnderutilizedPodsByCPUQuota(sysdigClient),
+		tools.NewTroubleshootKubernetesListUnderutilizedPodsByMemoryQuota(sysdigClient),
 	)
 	return handler
 }

internal/infra/mcp/tools/tool_troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota.go

@@ -0,0 +1,86 @@
+package tools
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"strings"
+
+	"github.com/mark3labs/mcp-go/mcp"
+	"github.com/mark3labs/mcp-go/server"
+	"github.com/sysdiglabs/sysdig-mcp-server/internal/infra/sysdig"
+)
+
+type TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota struct {
+	SysdigClient sysdig.ExtendedClientWithResponsesInterface
+}
+
+func NewTroubleshootKubernetesListUnderutilizedPodsByMemoryQuota(sysdigClient sysdig.ExtendedClientWithResponsesInterface) *TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota {
+	return &TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota{
+		SysdigClient: sysdigClient,
+	}
+}
+
+func (t *TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota) RegisterInServer(s *server.MCPServer) {
+	tool := mcp.NewTool("troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota",
+		mcp.WithDescription("List Kubernetes pods with memory usage below 25% of the limit."),
+		mcp.WithString("cluster_name", mcp.Description("The name of the cluster to filter by.")),
+		mcp.WithString("namespace_name", mcp.Description("The name of the namespace to filter by.")),
+		mcp.WithNumber("limit",
+			mcp.Description("Maximum number of pods to return."),
+			mcp.DefaultNumber(10),
+		),
+		mcp.WithOutputSchema[map[string]any](),
+		WithRequiredPermissions(), // FIXME(fede): Add the required permissions. It should be `promql.exec` but somehow the token does not have that permission even if you are able to execute queries.
+	)
+	s.AddTool(tool, t.handle)
+}
+
+func (t *TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota) handle(ctx context.Context, request mcp.CallToolRequest) (*mcp.CallToolResult, error) {
+	clusterName := mcp.ParseString(request, "cluster_name", "")
+	namespaceName := mcp.ParseString(request, "namespace_name", "")
+	limit := mcp.ParseInt(request, "limit", 10)
+
+	query := buildUnderutilizedPodsByMemoryQuery(clusterName, namespaceName)
+
+	limitQuery := sysdig.LimitQuery(limit)
+	params := &sysdig.GetQueryV1Params{
+		Query: query,
+		Limit: &limitQuery,
+	}
+
+	httpResp, err := t.SysdigClient.GetQueryV1(ctx, params)
+	if err != nil {
+		return mcp.NewToolResultErrorFromErr("failed to get underutilized pod list", err), nil
+	}
+
+	if httpResp.StatusCode != 200 {
+		bodyBytes, _ := io.ReadAll(httpResp.Body)
+		return mcp.NewToolResultErrorf("failed to get underutilized pod list: status code %d, body: %s", httpResp.StatusCode, string(bodyBytes)), nil
+	}
+
+	var queryResponse sysdig.QueryResponseV1
+	if err := json.NewDecoder(httpResp.Body).Decode(&queryResponse); err != nil {
+		return mcp.NewToolResultErrorFromErr("failed to decode response", err), nil
+	}
+
+	return mcp.NewToolResultJSON(queryResponse)
+}
+
+func buildUnderutilizedPodsByMemoryQuery(clusterName, namespaceName string) string {
+	filters := []string{}
+	if clusterName != "" {
+		filters = append(filters, fmt.Sprintf(`kube_cluster_name="%s"`, clusterName))
+	}
+	if namespaceName != "" {
+		filters = append(filters, fmt.Sprintf(`kube_namespace_name="%s"`, namespaceName))
+	}
+
+	filterString := ""
+	if len(filters) > 0 {
+		filterString = fmt.Sprintf("{%s}", strings.Join(filters, ","))
+	}
+
+	return fmt.Sprintf("sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_used_bytes%s) / (sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_limit_bytes%s) > 0) < 0.25", filterString, filterString)
+}

internal/infra/mcp/tools/tool_troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota_test.go

@@ -0,0 +1,86 @@
+package tools_test
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+
+	"github.com/mark3labs/mcp-go/mcp"
+	"github.com/mark3labs/mcp-go/server"
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+	"github.com/sysdiglabs/sysdig-mcp-server/internal/infra/mcp/tools"
+	"github.com/sysdiglabs/sysdig-mcp-server/internal/infra/sysdig"
+	"github.com/sysdiglabs/sysdig-mcp-server/internal/infra/sysdig/mocks"
+	"go.uber.org/mock/gomock"
+)
+
+var _ = Describe("TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota Tool", func() {
+	var (
+		tool       *tools.TroubleshootKubernetesListUnderutilizedPodsByMemoryQuota
+		mockSysdig *mocks.MockExtendedClientWithResponsesInterface
+		mcpServer  *server.MCPServer
+		ctrl       *gomock.Controller
+	)
+
+	BeforeEach(func() {
+		ctrl = gomock.NewController(GinkgoT())
+		mockSysdig = mocks.NewMockExtendedClientWithResponsesInterface(ctrl)
+		tool = tools.NewTroubleshootKubernetesListUnderutilizedPodsByMemoryQuota(mockSysdig)
+		mcpServer = server.NewMCPServer("test", "test")
+		tool.RegisterInServer(mcpServer)
+	})
+
+	It("should register successfully in the server", func() {
+		Expect(mcpServer.GetTool("troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota")).NotTo(BeNil())
+	})
+
+	When("listing underutilized pods", func() {
+		DescribeTable("it succeeds", func(ctx context.Context, toolName string, request mcp.CallToolRequest, expectedParamsRequested sysdig.GetQueryV1Params) {
+			mockSysdig.EXPECT().GetQueryV1(gomock.Any(), &expectedParamsRequested).Return(&http.Response{
+				StatusCode: http.StatusOK,
+				Body:       io.NopCloser(bytes.NewBufferString(`{"status":"success"}`)),
+			}, nil)
+
+			serverTool := mcpServer.GetTool(toolName)
+			result, err := serverTool.Handler(ctx, request)
+			Expect(err).NotTo(HaveOccurred())
+
+			resultData, ok := result.Content[0].(mcp.TextContent)
+			Expect(ok).To(BeTrue())
+			Expect(resultData.Text).To(MatchJSON(`{"status":"success"}`))
+		},
+			Entry(nil,
+				"troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota",
+				mcp.CallToolRequest{
+					Params: mcp.CallToolParams{
+						Name:      "troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota",
+						Arguments: map[string]any{},
+					},
+				},
+				sysdig.GetQueryV1Params{
+					Query: `sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_used_bytes) / (sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_limit_bytes) > 0) < 0.25`,
+					Limit: asPtr(sysdig.LimitQuery(10)),
+				},
+			),
+			Entry(nil,
+				"troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota",
+				mcp.CallToolRequest{
+					Params: mcp.CallToolParams{
+						Name: "troubleshoot_kubernetes_list_underutilized_pods_by_memory_quota",
+						Arguments: map[string]any{
+							"cluster_name":   "test-cluster",
+							"namespace_name": "test-namespace",
+							"limit":          20,
+						},
+					},
+				},
+				sysdig.GetQueryV1Params{
+					Query: `sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_used_bytes{kube_cluster_name="test-cluster",kube_namespace_name="test-namespace"}) / (sum by (kube_cluster_name, kube_namespace_name, kube_pod_name)(sysdig_container_memory_limit_bytes{kube_cluster_name="test-cluster",kube_namespace_name="test-namespace"}) > 0) < 0.25`,
+					Limit: asPtr(sysdig.LimitQuery(20)),
+				},
+			),
+		)
+	})
+})