test(perf): compare saved CLI startup benchmarks (#94812)

Summary:
- Adds saved CLI startup benchmark report comparison flags to `scripts/bench-cli-startup.ts`, plus JSON output coverage and changed-target routing expectations for the new test-helper importer.
- PR surface: Tests +77, Other +109. Total +186 across 4 files.
- Reproducibility: not applicable. as a feature/tooling PR. The prior PR defects were source-proven in review comments and the current head addresses them; I did not run local tests because this review was read-only.

Automerge notes:
- Ran the ClawSweeper repair loop before final review.
- Included post-review commit in the final squash: test(perf): compare saved CLI startup benchmarks

Validation:
- ClawSweeper review passed for head 1afa110f1b.
- Required merge gates passed before the squash merge.

Prepared head SHA: 1afa110f1b
Review: https://github.com/openclaw/openclaw/pull/94812#issuecomment-4748785428

Co-authored-by: clawsweeper <274271284+clawsweeper[bot]@users.noreply.github.com>
Co-authored-by: Felix Isaac Lim <38658663+FelixIsaac@users.noreply.github.com>

test/scripts/bench-cli-startup.test.ts

@@ -1,6 +1,9 @@
 // Bench Cli Startup tests cover bench cli startup script behavior.
+import { existsSync, readFileSync, writeFileSync } from "node:fs";
+import { join } from "node:path";
 import { describe, expect, it } from "vitest";
 import { testing } from "../../scripts/bench-cli-startup.ts";
+import { createTempDirTracker } from "../helpers/temp-dir.js";
 
 function withEnv<T>(env: Record<string, string | undefined>, callback: () => T): T {
   const previous = new Map<string, string | undefined>();
@@ -26,6 +29,72 @@ function withEnv<T>(env: Record<string, string | undefined>, callback: () => T):
 }
 
 describe("bench-cli-startup", () => {
+  it("writes compare-mode JSON output and creates parent directories", () => {
+    const tempDirs = createTempDirTracker();
+    const tmpDir = tempDirs.make("openclaw-cli-startup-compare-output-");
+    try {
+      const baselinePath = join(tmpDir, "baseline.json");
+      const candidatePath = join(tmpDir, "candidate.json");
+      const outputPath = join(tmpDir, "nested", "comparison.json");
+      const makeReport = (durationAvg: number, maxRssAvg: number) => ({
+        primary: {
+          entry: "openclaw.mjs",
+          cases: [
+            {
+              id: "version",
+              name: "--version",
+              args: ["--version"],
+              contract: null,
+              samples: [],
+              summary: {
+                sampleCount: 1,
+                durationMs: {
+                  avg: durationAvg,
+                  p50: durationAvg,
+                  p95: durationAvg,
+                  min: durationAvg,
+                  max: durationAvg,
+                },
+                firstOutputMs: null,
+                maxRssMb: {
+                  avg: maxRssAvg,
+                  p50: maxRssAvg,
+                  p95: maxRssAvg,
+                  min: maxRssAvg,
+                  max: maxRssAvg,
+                },
+                exitSummary: "code:0x1",
+              },
+            },
+          ],
+        },
+      });
+
+      writeFileSync(baselinePath, JSON.stringify(makeReport(100, 50)), "utf8");
+      writeFileSync(candidatePath, JSON.stringify(makeReport(125, 60)), "utf8");
+
+      const { comparison } = testing.readBenchmarkComparison(baselinePath, candidatePath);
+      testing.writeJsonOutput(outputPath, comparison);
+      expect(existsSync(outputPath)).toBe(true);
+      expect(JSON.parse(readFileSync(outputPath, "utf8"))).toEqual({
+        baseline: baselinePath,
+        candidate: candidatePath,
+        deltas: [
+          {
+            id: "version",
+            name: "--version",
+            durationAvgDeltaMs: 25,
+            durationAvgDeltaPct: 25,
+            maxRssAvgDeltaMb: 10,
+            maxRssAvgDeltaPct: 20,
+          },
+        ],
+      });
+    } finally {
+      tempDirs.cleanup();
+    }
+  });
+
   it("fails reports with no measured samples", () => {
     expect(
       testing.collectFailedSamples({

test/scripts/test-projects.test.ts

@@ -1918,6 +1918,8 @@ describe("scripts/test-projects changed-target routing", () => {
         "test/helpers/temp-dir.ts": "export const tempDir = 'x';\n",
         "test/helpers/temp-dir.test.ts":
           "import { tempDir } from './temp-dir.js';\nvoid tempDir;\n",
+        "test/scripts/bench-cli-startup.test.ts":
+          "import { tempDir } from '../helpers/temp-dir.js';\nvoid tempDir;\n",
         "src/foo.test.ts":
           "import { tempDir } from '../test/helpers/temp-dir.js';\nvoid tempDir;\n",
       },
@@ -1926,7 +1928,11 @@ describe("scripts/test-projects changed-target routing", () => {
       },
     );
 
-    expect(targets).toEqual(["test/helpers/temp-dir.test.ts", "src/foo.test.ts"]);
+    expect(targets).toEqual([
+      "test/helpers/temp-dir.test.ts",
+      "src/foo.test.ts",
+      "test/scripts/bench-cli-startup.test.ts",
+    ]);
   });
 
   it("keeps the broad changed run available for shared test helpers", () => {