test: replaced exact sleep assertions with tolerance-aware helper

- Added `expectSleepNear` to allow ±100ms variance on sleep duration checks.
- Updated `--thinking` test value from `"extended"` to `Effort.XHigh` enum constant.

packages/coding-agent/test/agent-session-python-cleanup.test.ts

@@ -1,4 +1,4 @@
-import { afterEach, beforeEach, describe, expect, it, vi } from "bun:test";
+import { afterEach, beforeEach, describe, expect, it, type Mock, vi } from "bun:test";
 import * as fs from "node:fs";
 import * as os from "node:os";
 import * as path from "node:path";
@@ -92,6 +92,15 @@ const mockPositiveSleepsImmediate = () => {
 		return realSleep(duration ?? 0);
 	});
 };
+
+const expectSleepNear = (sleepSpy: Mock<typeof Bun.sleep>, targetMs: number) => {
+	const minMs = targetMs - 100;
+	expect(
+		sleepSpy.mock.calls.some(
+			([duration]) => typeof duration === "number" && duration >= minMs && duration <= targetMs,
+		),
+	).toBe(true);
+};
 const createSession = async (
 	tempDir: string,
 	cwd: string,
@@ -415,7 +424,7 @@ describe("AgentSession python cleanup", () => {
 
 		const [toolResult] = await Promise.all([toolExecution, disposeSession]);
 
-		expect(sleepSpy).toHaveBeenCalledWith(3000);
+		expectSleepNear(sleepSpy, 3000);
 
 		expect(disposed).toBe(true);
 		expect(toolExecutionSettled).toBe(true);
@@ -460,7 +469,7 @@ describe("AgentSession python cleanup", () => {
 			firstDisposed = true;
 		});
 		await disposeFirst;
-		expect(sleepSpy).toHaveBeenCalledWith(3000);
+		expectSleepNear(sleepSpy, 3000);
 
 		expect(firstDisposed).toBe(true);
 		expect(firstExecutionSettled).toBe(false);
@@ -708,7 +717,7 @@ describe("AgentSession python cleanup", () => {
 		const sleepSpy = mockPositiveSleepsImmediate();
 
 		await session.dispose();
-		expect(sleepSpy).toHaveBeenCalledWith(3000);
+		expectSleepNear(sleepSpy, 3000);
 		const [firstResult, secondResult] = await Promise.all([firstExecution, secondExecution]);
 
 		expect(firstResult.cancelled).toBe(true);

packages/coding-agent/test/cli-hide-thinking-flag.test.ts

@@ -1,4 +1,5 @@
 import { describe, expect, it } from "bun:test";
+import { Effort } from "@oh-my-pi/pi-ai";
 import { parseArgs } from "../src/cli/args";
 
 describe("parseArgs — --hide-thinking flag", () => {
@@ -20,9 +21,9 @@ describe("parseArgs — --hide-thinking flag", () => {
 	});
 
 	it("parses --hide-thinking with --thinking flag (both can coexist)", () => {
-		const result = parseArgs(["--hide-thinking", "--thinking", "extended"]);
+		const result = parseArgs(["--hide-thinking", "--thinking", "xhigh"]);
 		expect(result.hideThinking).toBe(true);
-		expect(result.thinking).toBe("extended");
+		expect(result.thinking).toBe(Effort.XHigh);
 	});
 
 	it("parses --hide-thinking in any position", () => {