datadog-mcp 5.3.0 → 5.3.2

package/README.md

@@ -461,7 +461,8 @@ Note: Enrichment adds latency (fetches monitor list). Use for detailed investiga
 1. **Find errors in logs**: `logs({ action: "search", status: "error", sample: "diverse" })`
 2. **Extract trace_id** from log attributes (`dd.trace_id`)
 3. **Get full trace**: `traces({ action: "search", query: "trace_id:<id>" })`
-4. **Query APM metrics**: `metrics({ action: "query", query: "avg:trace.<service>.request.duration{*}" })`
+4. **Query APM metrics** (avg): `metrics({ action: "query", query: "avg:trace.express.request.duration{service:my-service}" })`
+5. **Query APM latency percentiles** (p95): `metrics({ action: "query", query: "p95:trace.express.request{service:my-service}" })` — note: use root metric without `.duration` suffix for percentiles
 
 ## Deep Links
 

package/dist/index.js

@@ -2572,10 +2572,21 @@ function registerMetricsTool(server, metricsV1Api, metricsV2Api, limits, site =
 - metadata: Get metric details (unit, type, description)
 
 APM METRICS (auto-generated from traces):
-- trace.{service}.hits - Request count
-- trace.{service}.errors - Error count
-- trace.{service}.duration - Latency (use avg:, p95:, max:)
-Example: max:trace.{service}.request.duration{*}`,
+Keyed by OPERATION name (e.g. express.request, pg.query), NOT service name.
+Filter by service using tags: {service:my-service}
+
+PERCENTILES (p50/p75/p90/p95/p99) \u2014 use the ROOT metric (distribution type):
+  p95:trace.express.request{service:my-service}
+
+AVG/SUM/MIN/MAX \u2014 use the .duration SUFFIX (pre-aggregated gauge):
+  avg:trace.express.request.duration{service:my-service}
+
+Other trace metrics (gauges):
+- trace.<operation>.hits - Request count
+- trace.<operation>.errors - Error count
+- trace.<operation>.apdex - Apdex score
+
+To discover operation names for a service, use: traces tool with action "services"`,
     InputSchema5,
     async ({ action, query, from, to, metric, limit, pointLimit }) => {
       try {
@@ -2923,7 +2934,7 @@ function registerTracesTool(server, spansApi, _servicesApi, limits, site = "data
   server.tool(
     "traces",
     `Analyze APM traces for request flow and latency debugging. Actions: search (find spans), aggregate (group stats), services (list APM services). Key filters: minDuration/maxDuration ("500ms", "2s"), httpStatus ("5xx", ">=400"), status (ok/error), errorMessage (grep).
-APM METRICS: Traces auto-generate metrics in trace.{service}.* namespace. Use metrics tool to query: avg:trace.{service}.request.duration{*}`,
+APM METRICS: Traces auto-generate metrics in trace.<operation>.* namespace (e.g. trace.express.request). Use metrics tool to query: avg:trace.express.request.duration{service:my-service}. For percentiles (p95), use the root metric WITHOUT .duration suffix: p95:trace.express.request{service:my-service}`,
     InputSchema6,
     async ({
       action,
@@ -3186,27 +3197,69 @@ function formatSlo(s) {
     timeframe: String(primaryThreshold?.timeframe ?? ""),
     tags: s.tags ?? [],
     status: {
-      // Note: SLI status requires a separate API call to getSLOHistory
       sli: null,
       errorBudgetRemaining: null,
       state: "unknown"
     },
+    overallStatus: [],
     createdAt: s.createdAt ? new Date(s.createdAt * 1e3).toISOString() : "",
     modifiedAt: s.modifiedAt ? new Date(s.modifiedAt * 1e3).toISOString() : ""
   };
 }
+function formatSearchSlo(slo) {
+  const attrs = slo.data?.attributes;
+  const primaryThreshold = attrs?.thresholds?.[0];
+  return {
+    id: slo.data?.id ?? "",
+    name: attrs?.name ?? "",
+    description: attrs?.description ?? null,
+    type: String(attrs?.sloType ?? "unknown"),
+    targetThreshold: primaryThreshold?.target ?? 0,
+    warningThreshold: primaryThreshold?.warning ?? null,
+    timeframe: String(primaryThreshold?.timeframe ?? ""),
+    tags: attrs?.allTags ?? [],
+    status: {
+      sli: attrs?.status?.sli ?? null,
+      errorBudgetRemaining: attrs?.status?.errorBudgetRemaining ?? null,
+      state: String(attrs?.status?.state ?? "unknown")
+    },
+    overallStatus: (attrs?.overallStatus ?? []).map((os) => ({
+      sli: os.status ?? null,
+      errorBudgetRemaining: os.errorBudgetRemaining ?? null,
+      state: String(os.state ?? "unknown"),
+      target: os.target ?? null,
+      timeframe: String(os.timeframe ?? "")
+    })),
+    createdAt: attrs?.createdAt ? new Date(attrs.createdAt * 1e3).toISOString() : "",
+    modifiedAt: attrs?.modifiedAt ? new Date(attrs.modifiedAt * 1e3).toISOString() : ""
+  };
+}
+function buildSearchQuery(query, tags) {
+  const parts = [];
+  if (query) parts.push(query);
+  if (tags?.length) parts.push(...tags);
+  return parts.join(" ");
+}
 async function listSlos(api, params, limits) {
   const effectiveLimit = params.limit ?? limits.defaultLimit;
-  const response = await api.listSLOs({
-    ids: params.ids?.join(","),
-    query: params.query,
-    tagsQuery: params.tags?.join(","),
-    limit: effectiveLimit
+  if (params.ids?.length) {
+    const response2 = await api.listSLOs({
+      ids: params.ids.join(","),
+      limit: effectiveLimit
+    });
+    const slos3 = (response2.data ?? []).map(formatSlo);
+    return { slos: slos3, total: slos3.length };
+  }
+  const searchQuery = buildSearchQuery(params.query, params.tags);
+  const response = await api.searchSLO({
+    query: searchQuery || void 0,
+    pageSize: effectiveLimit
   });
-  const slos2 = (response.data ?? []).map(formatSlo);
+  const searchSlos = response.data?.attributes?.slos ?? [];
+  const slos2 = searchSlos.map(formatSearchSlo);
   return {
     slos: slos2,
-    total: response.data?.length ?? 0
+    total: slos2.length
   };
 }
 async function getSlo(api, id) {
@@ -3298,7 +3351,7 @@ async function getSloHistory(api, id, params) {
 function registerSlosTool(server, api, limits, readOnly = false, _site = "datadoghq.com") {
   server.tool(
     "slos",
-    "Manage Datadog Service Level Objectives. Actions: list, get, create, update, delete, history. SLO types: metric-based, monitor-based. Use for: reliability tracking, error budgets, SLA compliance, performance targets.",
+    "Manage Datadog Service Level Objectives. Actions: list (with SLI status & error budget), get, create, update, delete, history. SLO types: metric-based, monitor-based. Use for: reliability tracking, error budgets, SLA compliance, performance targets.",
     InputSchema8,
     async ({ action, id, ids, query, tags, limit, config, from, to }) => {
       try {