diff --git a/packages/mcp-core/src/api-client/schema.ts b/packages/mcp-core/src/api-client/schema.ts
index cccc58224..e5b80b098 100644
--- a/packages/mcp-core/src/api-client/schema.ts
+++ b/packages/mcp-core/src/api-client/schema.ts
@@ -2447,7 +2447,11 @@ export const EventsStatsResponseSchema = z
     data: z.array(
       z.tuple([
         z.number(),
-        z.array(z.object({ count: z.number().nullish() }).passthrough()),
+        z.array(
+          z
+            .object({ count: z.union([z.string(), z.number()]).nullish() })
+            .passthrough(),
+        ),
       ]),
     ),
     start: z.number().optional(),
diff --git a/packages/mcp-core/src/tools/catalog/search-events.test.ts b/packages/mcp-core/src/tools/catalog/search-events.test.ts
index 7de937048..44d0e042d 100644
--- a/packages/mcp-core/src/tools/catalog/search-events.test.ts
+++ b/packages/mcp-core/src/tools/catalog/search-events.test.ts
@@ -306,6 +306,137 @@ describe("search_events", () => {
     expect(result).toContain("**Peak**: 8");
   });
 
+  it("handles timeseries when the API returns count as a numeric string (e.g. max(timestamp) epoch)", async () => {
+    const output = {
+      dataset: "errors" as const,
+      query: "",
+      fields: [] as string[],
+      sort: "-timestamp",
+      environment: null,
+      timeSeries: { yAxis: "max(timestamp)", interval: "1d" },
+      timeRange: { statsPeriod: "45d" },
+      explanation: "Max timestamp per day",
+    };
+    mockGenerateText.mockResolvedValueOnce({
+      text: JSON.stringify(output),
+      experimental_output: output,
+      finishReason: "stop" as const,
+      usage: { promptTokens: 10, completionTokens: 5, totalTokens: 15 },
+      warnings: [] as const,
+    } as any);
+
+    // The Sentry API returns `count` as a string for non-count aggregates; numeric
+    // epoch strings should be parsed correctly and never render as "NaN".
+    mswServer.use(
+      http.get(
+        "https://sentry.io/api/0/organizations/test-org/events-stats/",
+        () =>
+          HttpResponse.json({
+            data: [
+              [1757548800, [{ count: "1757548800" }]],
+              [1757635200, [{ count: "1757635200" }]],
+              [1757721600, [{ count: "1757721600" }]],
+            ],
+          }),
+      ),
+    );
+
+    const result = await searchEvents.handler(
+      {
+        organizationSlug: "test-org",
+        regionUrl: null,
+        projectSlug: null,
+        dataset: "errors",
+        query: "latest event timestamp per day",
+        fields: null,
+        sort: null,
+        period: "45d",
+        limit: 10,
+        includeExplanation: false,
+      },
+      {
+        accessToken: "test-token",
+        userId: "user-123",
+        clientId: "client-123",
+        grantedSkills: new Set(),
+        constraints: {},
+        sentryHost: "sentry.io",
+      },
+    );
+
+    expect(result).toContain("max(timestamp) over time");
+    expect(result).not.toContain("NaN");
+  });
+
+  it("handles timeseries when the API returns count as an ISO datetime string", async () => {
+    const output = {
+      dataset: "errors" as const,
+      query: "",
+      fields: [] as string[],
+      sort: "-timestamp",
+      environment: null,
+      timeSeries: { yAxis: "max(timestamp)", interval: "1d" },
+      timeRange: { statsPeriod: "7d" },
+      explanation: "Max timestamp per day",
+    };
+    mockGenerateText.mockResolvedValueOnce({
+      text: JSON.stringify(output),
+      experimental_output: output,
+      finishReason: "stop" as const,
+      usage: { promptTokens: 10, completionTokens: 5, totalTokens: 15 },
+      warnings: [] as const,
+    } as any);
+
+    // When the API returns ISO datetime strings, the output must display the raw
+    // string rather than "NaN".
+    mswServer.use(
+      http.get(
+        "https://sentry.io/api/0/organizations/test-org/events-stats/",
+        () =>
+          HttpResponse.json({
+            // Empty buckets are zero-filled with a numeric 0 by Sentry.
+            data: [
+              [1757462400, [{ count: 0 }]],
+              [1757548800, [{ count: "2026-07-11T00:00:00+00:00" }]],
+              [1757635200, [{ count: "2026-07-12T00:00:00+00:00" }]],
+              [1757721600, [{ count: "2026-07-13T00:00:00+00:00" }]],
+            ],
+          }),
+      ),
+    );
+
+    const result = await searchEvents.handler(
+      {
+        organizationSlug: "test-org",
+        regionUrl: null,
+        projectSlug: null,
+        dataset: "errors",
+        query: "latest event timestamp per day",
+        fields: null,
+        sort: null,
+        period: "7d",
+        limit: 10,
+        includeExplanation: false,
+      },
+      {
+        accessToken: "test-token",
+        userId: "user-123",
+        clientId: "client-123",
+        grantedSkills: new Set(),
+        constraints: {},
+        sentryHost: "sentry.io",
+      },
+    );
+
+    expect(result).toContain("max(timestamp) over time");
+    // ISO datetime strings must appear in the output as-is, never as "NaN".
+    expect(result).toContain("2026-07-13T00:00:00+00:00");
+    expect(result).toContain(
+      "**Peak**: 2026-07-13T00:00:00+00:00 at 2025-09-13 00:00",
+    );
+    expect(result).not.toContain("NaN");
+  });
+
   it("should handle spans dataset queries", async () => {
     // Mock AI response for spans dataset
     mockGenerateText.mockResolvedValueOnce(
diff --git a/packages/mcp-core/src/tools/support/search-events/formatters.ts b/packages/mcp-core/src/tools/support/search-events/formatters.ts
index 624559bde..4796e019a 100644
--- a/packages/mcp-core/src/tools/support/search-events/formatters.ts
+++ b/packages/mcp-core/src/tools/support/search-events/formatters.ts
@@ -943,7 +943,7 @@ function isAdditiveAggregate(yAxis: string): boolean {
  * `interval` is null when Sentry chose the bucket size for the range.
  */
 export function formatTimeSeriesResults(params: {
-  series: { data: Array<[number, Array<{ count?: number | null }>]> };
+  series: { data: Array<[number, Array<{ count?: string | number | null }>]> };
   yAxis: string;
   interval: string | null;
   inputQuery: string;
@@ -963,10 +963,21 @@ export function formatTimeSeriesResults(params: {
     url,
   } = params;
 
-  const points = series.data.map(([ts, values]) => ({
-    time: new Date(ts * 1000).toISOString().slice(0, 16).replace("T", " "),
-    value: values[0]?.count ?? 0,
-  }));
+  const points = series.data.map(([ts, values]) => {
+    const raw = values[0]?.count ?? 0;
+    // Date-typed aggregates such as max(timestamp) return ISO datetime
+    // strings. Number() rejects those whole (parseFloat would keep the year),
+    // so order them by parsed time and display the raw string.
+    const numeric = typeof raw === "number" ? raw : Number(raw);
+    const isNumeric = !Number.isNaN(numeric);
+    const value = isNumeric ? numeric : Date.parse(String(raw)) || 0;
+    const display = isNumeric ? numeric.toLocaleString() : String(raw);
+    return {
+      time: new Date(ts * 1000).toISOString().slice(0, 16).replace("T", " "),
+      value,
+      display,
+    };
+  });
 
   // Total is only meaningful for additive aggregates; summing count_unique /
   // avg / percentile buckets would be wrong, so omit it for those.
@@ -1000,7 +1011,7 @@ export function formatTimeSeriesResults(params: {
     lines.push(`- **Total**: ${total.toLocaleString()}`);
   }
   if (peak) {
-    lines.push(`- **Peak**: ${peak.value.toLocaleString()} at ${peak.time}`);
+    lines.push(`- **Peak**: ${peak.display} at ${peak.time}`);
   }
 
   if (shown.length > 0) {
@@ -1012,7 +1023,7 @@ export function formatTimeSeriesResults(params: {
       "| --- | --- |",
     );
     for (const p of shown) {
-      lines.push(`| ${p.time} | ${p.value.toLocaleString()} |`);
+      lines.push(`| ${p.time} | ${p.display} |`);
     }
   } else {
     lines.push("", "No data points in this range.");