fix(test-harness): restore promisify(exec) + unblock CLI introspection probes

The vitest child-process guard wrapped exec/execFile without preserving the
`[util.promisify.custom]` symbol, so awaited `execAsync` resolved to a raw
stdout string instead of `{stdout, stderr}`. That single regression cascaded
through ~60 "failing" tests across cli, core, engine, and dashboard whose
production code was actually correct. Also relax the AI-CLI blocklist for
cheap introspection (--version/--help/which …), give SIGTERM'd subprocesses a
brief grace period before being flagged as "left running", fix a few real
test-side bugs uncovered along the way (executor mock step transitions, iOS
last-resort keyboard path, mission SSE replay tests racing with the real AI
agent), and convert dashboard route tests' dynamic `await import("../server.js")`
to static imports so first-test timings drop from 2–5s to <200ms.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

packages/core/src/__test-utils__/vitest-setup.ts

@@ -16,6 +16,7 @@ import { afterEach, expect } from "vitest";
 import { createRequire, syncBuiltinESMExports } from "node:module";
 import { tmpdir } from "node:os";
 import { dirname, join, resolve } from "node:path";
+import { promisify } from "node:util";
 import { isMainThread } from "node:worker_threads";
 import { assertOutsideRealFusionPath } from "../test-safety.js";
 
@@ -402,11 +403,36 @@ function currentTestName(): string | null {
   return expect.getState().currentTestName ?? null;
 }
 
+// Cheap, no-network introspection invocations are safe to run in tests — they
+// don't open an AI session, don't hit a paid API, and the dashboard's CLI
+// availability probe needs them to tell the truth about the local system.
+//
+// This must stay strict: only exact "is this binary installed / what version is
+// it?" probes are allowed. Do not match `--help` / `--version` substrings
+// inside arbitrary prompt text, or the test guard can be bypassed.
+const SAFE_INTROSPECTION_LOOKUP_PATTERN =
+  /^\s*(?:which|where|type)\s+(?:-[a-zA-Z]+\s+)*(?:"[^"]+"|'[^']+'|\S+)\s*$/i;
+const SAFE_INTROSPECTION_COMMAND_V_PATTERN =
+  /^\s*command\s+-v\s+(?:"[^"]+"|'[^']+'|\S+)\s*$/i;
+const SAFE_INTROSPECTION_BLOCKED_CLI_PATTERN =
+  /^\s*(?:"[^"]*(?:claude|droid|paperclipai|hermes|openclaw)(?:\.(?:cmd|bat|ps1|exe))?[^"]*"|'[^']*(?:claude|droid|paperclipai|hermes|openclaw)(?:\.(?:cmd|bat|ps1|exe))?[^']*'|(?:\S+[\\/])?(?:claude|droid|paperclipai|hermes|openclaw)(?:\.(?:cmd|bat|ps1|exe))?)\s+(?:--version|--help|-V|-h)\s*$/i;
+
+function isSafeIntrospectionCommand(commandLine: string): boolean {
+  return (
+    SAFE_INTROSPECTION_LOOKUP_PATTERN.test(commandLine) ||
+    SAFE_INTROSPECTION_COMMAND_V_PATTERN.test(commandLine) ||
+    SAFE_INTROSPECTION_BLOCKED_CLI_PATTERN.test(commandLine)
+  );
+}
+
 function shouldBlockRealTestCli(commandLine: string): boolean {
   if (process.env.FUSION_TEST_ALLOW_REAL_AI_CLI === "1") {
     return false;
   }
-  return BLOCKED_TEST_CLI_PATTERN.test(commandLine);
+  if (!BLOCKED_TEST_CLI_PATTERN.test(commandLine)) {
+    return false;
+  }
+  return !isSafeIntrospectionCommand(commandLine);
 }
 
 function blockedCliError(commandLine: string): Error {
@@ -509,7 +535,10 @@ function installChildProcessGuards(): void {
     return originalChildProcess.execFileSync(file, args, options);
   }) as ChildProcessModule["execFileSync"];
 
-  mutableChildProcess.exec = ((command: string, optionsOrCallback?: ExecOptions | ((error: Error | null, stdout: string, stderr: string) => void), maybeCallback?: (error: Error | null, stdout: string, stderr: string) => void) => {
+  // Preserve util.promisify(exec) → { stdout, stderr } semantics. Function.prototype.bind
+  // and our wrapper drop the original [util.promisify.custom] symbol, which would otherwise
+  // make awaited execAsync resolve to a raw stdout string and break destructuring.
+  const execWrapper = ((command: string, optionsOrCallback?: ExecOptions | ((error: Error | null, stdout: string, stderr: string) => void), maybeCallback?: (error: Error | null, stdout: string, stderr: string) => void) => {
     if (shouldBlockRealTestCli(command)) {
       throw blockedCliError(command);
     }
@@ -518,9 +547,22 @@ function installChildProcessGuards(): void {
     const proc = originalChildProcess.exec(command, withDefaultTimeout(options), callback);
     registerTrackedSubprocess(proc, command);
     return proc;
-  }) as ChildProcessModule["exec"];
+  }) as unknown as ChildProcessModule["exec"];
+  (execWrapper as unknown as Record<symbol, unknown>)[promisify.custom] = (command: string, options?: ExecOptions) =>
+    new Promise<{ stdout: string; stderr: string }>((resolve, reject) => {
+      execWrapper(command, options ?? {}, (error: Error | null, stdout: string, stderr: string) => {
+        if (error) {
+          (error as Error & { stdout?: string; stderr?: string }).stdout = stdout;
+          (error as Error & { stdout?: string; stderr?: string }).stderr = stderr;
+          reject(error);
+          return;
+        }
+        resolve({ stdout, stderr });
+      });
+    });
+  mutableChildProcess.exec = execWrapper;
 
-  mutableChildProcess.execFile = ((file: string, argsOrOptions?: readonly string[] | ExecFileOptions | ((error: Error | null, stdout: string, stderr: string) => void), optionsOrCallback?: ExecFileOptions | ((error: Error | null, stdout: string, stderr: string) => void), maybeCallback?: (error: Error | null, stdout: string, stderr: string) => void) => {
+  const execFileWrapper = ((file: string, argsOrOptions?: readonly string[] | ExecFileOptions | ((error: Error | null, stdout: string, stderr: string) => void), optionsOrCallback?: ExecFileOptions | ((error: Error | null, stdout: string, stderr: string) => void), maybeCallback?: (error: Error | null, stdout: string, stderr: string) => void) => {
     const args = Array.isArray(argsOrOptions) ? [...argsOrOptions] : [];
     const commandLine = describeTestSubprocessCommand(file, args);
     if (shouldBlockRealTestCli(commandLine)) {
@@ -535,7 +577,20 @@ function installChildProcessGuards(): void {
     const proc = originalChildProcess.execFile(file, args, withDefaultTimeout(options), callback);
     registerTrackedSubprocess(proc, commandLine);
     return proc;
-  }) as ChildProcessModule["execFile"];
+  }) as unknown as ChildProcessModule["execFile"];
+  (execFileWrapper as unknown as Record<symbol, unknown>)[promisify.custom] = (file: string, args?: readonly string[], options?: ExecFileOptions) =>
+    new Promise<{ stdout: string; stderr: string }>((resolve, reject) => {
+      execFileWrapper(file, args ?? [], options ?? {}, (error: Error | null, stdout: string, stderr: string) => {
+        if (error) {
+          (error as Error & { stdout?: string; stderr?: string }).stdout = stdout;
+          (error as Error & { stdout?: string; stderr?: string }).stderr = stderr;
+          reject(error);
+          return;
+        }
+        resolve({ stdout, stderr });
+      });
+    });
+  mutableChildProcess.execFile = execFileWrapper;
 
   mutableChildProcess.fork = ((modulePath: string, argsOrOptions?: readonly string[] | ForkOptions, maybeOptions?: ForkOptions) => {
     const args = Array.isArray(argsOrOptions) ? [...argsOrOptions] : [];
@@ -554,10 +609,48 @@ function installChildProcessGuards(): void {
 
 installChildProcessGuards();
 
-afterEach(() => {
+afterEach(async () => {
   const failures = [...completedSubprocessFailures];
   completedSubprocessFailures.length = 0;
 
+  // Give SIGTERM'd processes a brief grace period to exit before declaring
+  // them "left running" — tests like dev-server-process.cleanup() send SIGTERM
+  // and immediately drop their reference, so the OS exit lags the test by a
+  // few ms even when the production code did the right thing.
+  const SUBPROCESS_GRACE_MS = 200;
+  if (trackedSubprocesses.size > 0) {
+    const stillRunningProcs: ChildProcess[] = [];
+    for (const [proc] of trackedSubprocesses) {
+      if (proc.exitCode === null && proc.signalCode === null) {
+        stillRunningProcs.push(proc);
+      }
+    }
+    if (stillRunningProcs.length > 0) {
+      await new Promise<void>((resolve) => {
+        let remaining = stillRunningProcs.length;
+        const done = () => {
+          remaining -= 1;
+          if (remaining <= 0) resolve();
+        };
+        const timer = setTimeout(() => resolve(), SUBPROCESS_GRACE_MS);
+        for (const proc of stillRunningProcs) {
+          if (proc.exitCode !== null || proc.signalCode !== null) {
+            done();
+            continue;
+          }
+          const finish = () => {
+            proc.removeListener("exit", finish);
+            proc.removeListener("close", finish);
+            done();
+          };
+          proc.once("exit", finish);
+          proc.once("close", finish);
+        }
+        timer.unref?.();
+      });
+    }
+  }
+
   for (const [proc, tracked] of trackedSubprocesses) {
     const stillRunning = proc.exitCode === null && proc.signalCode === null;
     if (stillRunning) {

packages/core/src/__tests__/setup-test-isolation.test.ts

@@ -86,10 +86,34 @@ describe("test isolation setup", () => {
   });
 
   it("blocks real AI CLI subprocesses from running in tests", () => {
-    expect(() => spawnSync("droid", ["--version"])).toThrow(
+    // Interactive / session-launching invocations stay blocked.
+    expect(() => spawnSync("droid", ["chat"])).toThrow(
       "Real AI CLI launch blocked during tests",
     );
-    expect(() => execSync("claude --version", { encoding: "utf-8" })).toThrow(
+    expect(() => execSync("claude -p 'hello world'", { encoding: "utf-8" })).toThrow(
+      "Real AI CLI launch blocked during tests",
+    );
+  });
+
+  it("permits cheap introspection invocations (--version, --help)", () => {
+    // These probe whether the binary is installed without opening an AI
+    // session, and the dashboard CLI-availability probe needs them. We use
+    // binaries from the blocklist that are not installed on test runners
+    // (`openclaw`, `paperclipai`) so the spawn returns ENOENT quickly rather
+    // than actually launching a real CLI on a developer machine.
+    expect(() => spawnSync("openclaw", ["--version"])).not.toThrow(
+      "Real AI CLI launch blocked during tests",
+    );
+    expect(() => spawnSync("paperclipai", ["--help"])).not.toThrow(
+      "Real AI CLI launch blocked during tests",
+    );
+  });
+
+  it("keeps blocking prompt invocations that merely mention --version/--help", () => {
+    expect(() => execSync('claude -p "please print --version literally"', { encoding: "utf-8" })).toThrow(
+      "Real AI CLI launch blocked during tests",
+    );
+    expect(() => spawnSync("openclaw", ["-p", "say --help literally"])).toThrow(
       "Real AI CLI launch blocked during tests",
     );
   });

packages/dashboard/app/hooks/__tests__/useMobileKeyboard.test.ts

@@ -288,6 +288,9 @@ describe("useMobileKeyboard", () => {
   });
 
   it("treats focused input + viewport shrink as keyboard-open even when overlap is 0", async () => {
+    // iOS last-resort path: chromeOverlap = 0 (innerHeight tracks offsetTop+vv.height),
+    // gap < 16 (focused-fallback doesn't fire), and viewportShrink >= 16 from the
+    // baseline so the focused-input shrink heuristic is the only signal left.
     const { listeners, mockVV } = setupMobileVisualViewport({
       innerHeight: 844,
       vvHeight: 844,
@@ -305,7 +308,17 @@ describe("useMobileKeyboard", () => {
 
     input.focus();
     Object.defineProperty(mockVV, "height", {
-      value: 826,
+      value: 824,
+      writable: true,
+      configurable: true,
+    });
+    Object.defineProperty(mockVV, "offsetTop", {
+      value: 5,
+      writable: true,
+      configurable: true,
+    });
+    Object.defineProperty(window, "innerHeight", {
+      value: 829,
       writable: true,
       configurable: true,
     });
@@ -316,7 +329,7 @@ describe("useMobileKeyboard", () => {
 
     await waitFor(() => {
       expect(result.current.keyboardOverlap).toBe(0);
-      expect(result.current.viewportHeight).toBe(826);
+      expect(result.current.viewportHeight).toBe(824);
       expect(result.current.keyboardOpen).toBe(true);
     });
 

packages/dashboard/src/__tests__/mission-e2e.test.ts

@@ -2212,7 +2212,8 @@ describe("Mission API", () => {
 
     it("replays buffered interview events when Last-Event-ID is provided", async () => {
       const { app } = buildApp();
-      const sessionId = await createMissionInterviewSession("127.0.0.1", "Replay Mission", "/tmp/project");
+      const sessionId = "replay-test-session";
+      missionInterviewModule.__registerMissionInterviewSessionForTest(sessionId, "Replay Mission");
 
       missionInterviewStreamManager.broadcast(sessionId, { type: "thinking", data: "first" });
       missionInterviewStreamManager.broadcast(sessionId, { type: "thinking", data: "second" });
@@ -2239,7 +2240,8 @@ describe("Mission API", () => {
 
     it("does not replay buffered interview events when Last-Event-ID is missing", async () => {
       const { app } = buildApp();
-      const sessionId = await createMissionInterviewSession("127.0.0.1", "No Replay Mission", "/tmp/project");
+      const sessionId = "no-replay-test-session";
+      missionInterviewModule.__registerMissionInterviewSessionForTest(sessionId, "No Replay Mission");
 
       missionInterviewStreamManager.broadcast(sessionId, { type: "thinking", data: "first" });
 
@@ -2261,7 +2263,8 @@ describe("Mission API", () => {
 
     it("gracefully ignores invalid Last-Event-ID values for interview streams", async () => {
       const { app } = buildApp();
-      const sessionId = await createMissionInterviewSession("127.0.0.1", "Invalid Replay Mission", "/tmp/project");
+      const sessionId = "invalid-replay-test-session";
+      missionInterviewModule.__registerMissionInterviewSessionForTest(sessionId, "Invalid Replay Mission");
 
       missionInterviewStreamManager.broadcast(sessionId, { type: "thinking", data: "first" });
 

packages/dashboard/src/__tests__/routes-agent-budget.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { get, request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 // ── Mock @fusion/core for budget routes ─────────────────────────────────
 
@@ -68,9 +69,9 @@ function createMockBudgetStatus(overrides: Record<string, unknown> = {}) {
 
 describe("Agent budget routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
-  beforeEach(async () => {
+  beforeEach(() => {
     vi.clearAllMocks();
     mockInit.mockResolvedValue(undefined);
     mockGetAgent.mockResolvedValue({ id: "agent-001", state: "running" });
@@ -78,7 +79,6 @@ describe("Agent budget routes", () => {
     mockResetBudgetUsage.mockResolvedValue(undefined);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-export.test.ts

@@ -5,6 +5,7 @@ import { join } from "node:path";
 import { tmpdir } from "node:os";
 
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 const mockInit = vi.fn().mockResolvedValue(undefined);
 const mockListAgents = vi.fn().mockResolvedValue([]);
@@ -53,7 +54,7 @@ async function postExport(app: Parameters<typeof request>[0], body: unknown) {
 
 describe("POST /api/agents/export", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
   let testDir: string;
 
   beforeEach(async () => {
@@ -91,7 +92,6 @@ describe("POST /api/agents/export", () => {
     });
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-import.test.ts

@@ -4,6 +4,7 @@ import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from "node:fs";
 import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 const mockInit = vi.fn().mockResolvedValue(undefined);
 const mockListAgents = vi.fn().mockResolvedValue([]);
@@ -16,12 +17,21 @@ const mockParseSingleAgentManifest = vi.fn();
 const mockPrepareAgentCompaniesImport = vi.fn();
 
 // Use vi.hoisted to ensure mocks are available when vi.mock runs
-const { mockFsAccess, mockFsMkdir, mockFsWriteFile, mockExecFile } = vi.hoisted(() => ({
-  mockFsAccess: vi.fn(),
-  mockFsMkdir: vi.fn(),
-  mockFsWriteFile: vi.fn(),
-  mockExecFile: vi.fn(),
-}));
+const { mockFsAccess, mockFsMkdir, mockFsWriteFile, mockExecFile, MockAgentCompaniesParseError } = vi.hoisted(() => {
+  class MockAgentCompaniesParseError extends Error {
+    constructor(message: string) {
+      super(message);
+      this.name = "AgentCompaniesParseError";
+    }
+  }
+  return {
+    mockFsAccess: vi.fn(),
+    mockFsMkdir: vi.fn(),
+    mockFsWriteFile: vi.fn(),
+    mockExecFile: vi.fn(),
+    MockAgentCompaniesParseError,
+  };
+});
 
 vi.mock("node:fs/promises", async (importOriginal) => {
   const actual = await importOriginal() as Record<string, unknown>;
@@ -47,13 +57,6 @@ vi.mock("node:child_process", async (importOriginal) => {
   };
 });
 
-class MockAgentCompaniesParseError extends Error {
-  constructor(message: string) {
-    super(message);
-    this.name = "AgentCompaniesParseError";
-  }
-}
-
 vi.mock("@fusion/core", () => {
   return {
     AgentStore: class MockAgentStore {
@@ -107,7 +110,7 @@ const originalFetch = globalThis.fetch;
 
 describe("POST /api/agents/import", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
   let testDir: string;
 
   beforeEach(async () => {
@@ -175,7 +178,6 @@ describe("POST /api/agents/import", () => {
     });
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 
@@ -869,7 +871,7 @@ describe("POST /api/agents/import", () => {
 
 describe("GET /api/agents/companies", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
   const originalFetch = globalThis.fetch;
 
   beforeEach(async () => {
@@ -882,7 +884,6 @@ describe("GET /api/agents/companies", () => {
     mockInit.mockResolvedValue(undefined);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-keys.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 const mockInit = vi.fn().mockResolvedValue(undefined);
 const mockGetAgent = vi.fn();
@@ -49,7 +50,7 @@ class MockStore extends EventEmitter {
 
 describe("Agent API key routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
   beforeEach(async () => {
     vi.clearAllMocks();
@@ -57,7 +58,6 @@ describe("Agent API key routes", () => {
     mockListAgents.mockResolvedValue([]);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-ratings.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 // ── Mock @fusion/core for agent ratings ─────────────────────────────────
 
@@ -83,14 +84,13 @@ function createMockSummary(overrides: Record<string, unknown> = {}) {
 
 describe("Agent ratings routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
   beforeEach(async () => {
     vi.clearAllMocks();
     mockInit.mockResolvedValue(undefined);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-revisions.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 const mockInit = vi.fn().mockResolvedValue(undefined);
 const mockGetAgent = vi.fn();
@@ -83,7 +84,7 @@ function createMockRevision(overrides: Record<string, unknown> = {}) {
 
 describe("Agent config revision routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
   beforeEach(async () => {
     vi.clearAllMocks();
@@ -91,7 +92,6 @@ describe("Agent config revision routes", () => {
     mockListAgents.mockResolvedValue([]);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-agent-soul-memory.test.ts

@@ -5,6 +5,7 @@ import { mkdir, rm, writeFile } from "node:fs/promises";
 import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 type AgentRecord = {
   id: string;
@@ -84,7 +85,7 @@ function createAgent(overrides: Partial<AgentRecord> = {}): AgentRecord {
 
 describe("Agent soul/memory routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
   let agents: Map<string, AgentRecord>;
   let tempDir: string;
 
@@ -123,7 +124,6 @@ describe("Agent soul/memory routes", () => {
     await mkdir(join(tempDir, ".fusion"), { recursive: true });
 
     store = new MockStore(tempDir);
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-nodes-sync.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request, get } from "../test-request.js";
+import { createServer } from "../server.js";
 import { resetRuntimeLogSink, setRuntimeLogSink, type RuntimeLogContext } from "../runtime-logger.js";
 
 // Mock node:fs for auth.json reading
@@ -157,7 +158,7 @@ interface RuntimeEvent {
 
 describe("Node settings sync routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
   let mockFetch: ReturnType<typeof vi.fn>;
   let runtimeEvents: RuntimeEvent[];
 
@@ -185,7 +186,6 @@ describe("Node settings sync routes", () => {
     });
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-nodes.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request, get } from "../test-request.js";
+import { createServer } from "../server.js";
 
 // ── Mock @fusion/core for node routes ─────────────────────────────────
 
@@ -85,7 +86,7 @@ function createMockNode(overrides: Record<string, unknown> = {}) {
 
 describe("Node routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
   beforeEach(async () => {
     vi.clearAllMocks();
@@ -101,7 +102,6 @@ describe("Node routes", () => {
     mockGetDiscoveryConfig.mockReturnValue(null);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-org-chart.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { request } from "../test-request.js";
+import { createServer } from "../server.js";
 
 const mockInit = vi.fn().mockResolvedValue(undefined);
 const mockGetAgent = vi.fn();
@@ -62,7 +63,7 @@ function createMockAgent(id: string, name: string, reportsTo?: string) {
 
 describe("Agent org chart routes", () => {
   let store: MockStore;
-  let app: ReturnType<typeof import("../server.js").createServer>;
+  let app: ReturnType<typeof createServer>;
 
   beforeEach(async () => {
     vi.clearAllMocks();
@@ -70,7 +71,6 @@ describe("Agent org chart routes", () => {
     mockListAgents.mockResolvedValue([]);
 
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as any);
   });
 

packages/dashboard/src/__tests__/routes-projects-across-nodes.test.ts

@@ -1,6 +1,7 @@
 import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
 import { EventEmitter } from "node:events";
 import { get } from "../test-request.js";
+import { createServer } from "../server.js";
 
 // ── Mock @fusion/core for project routes ─────────────────────────────────
 
@@ -109,10 +110,9 @@ describe("GET /api/projects/across-nodes", () => {
   let store: MockStore;
   let app: (req: import("node:http").IncomingMessage, res: import("node:http").ServerResponse) => void;
 
-  beforeEach(async () => {
+  beforeEach(() => {
     vi.clearAllMocks();
     store = new MockStore();
-    const { createServer } = await import("../server.js");
     app = createServer(store as unknown as Parameters<typeof createServer>[0] extends { store: infer S } ? S : never);
   });
 

packages/dashboard/src/mission-interview.ts

@@ -1263,6 +1263,25 @@ export function cleanupMissionInterviewSession(sessionId: string): void {
 /**
  * Reset all mission interview state. Used for testing only.
  */
+/**
+ * Test-only: register a stub session so SSE/buffer tests can drive the stream
+ * manager directly without spinning up the real AI agent (which is
+ * deliberately blocked in the vitest harness).
+ */
+export function __registerMissionInterviewSessionForTest(sessionId: string, missionTitle = "Test Mission"): void {
+  sessions.set(sessionId, {
+    id: sessionId,
+    ip: "127.0.0.1",
+    missionId: "",
+    missionTitle,
+    history: [],
+    thinkingOutput: "",
+    lastGeneratedThinking: "",
+    createdAt: new Date(),
+    updatedAt: new Date(),
+  });
+}
+
 export function __resetMissionInterviewState(): void {
   for (const [id] of sessions) {
     cleanupInMemoryMissionSession(id);

packages/engine/src/__tests__/executor.test.ts

@@ -4406,12 +4406,21 @@ const mockedReviewStep = vi.mocked(mockedReviewStepFn);
  */
 async function captureTools(): Promise<Record<string, (id: string, params: any) => Promise<any>>> {
   const store = createMockStore();
-  store.updateStep.mockResolvedValue({
-    steps: [
-      { name: "Preflight", status: "done" },
-      { name: "Implement", status: "in-progress" },
-      { name: "Testing", status: "pending" },
-    ],
+  // Simulate the real TaskStore: forward transitions persist, but in-progress
+  // regressions on done/skipped steps are rejected so executor.ts can surface
+  // the "already <status>" diagnostic.
+  const stepStates: Array<{ name: string; status: string }> = [
+    { name: "Preflight", status: "done" },
+    { name: "Implement", status: "in-progress" },
+    { name: "Testing", status: "pending" },
+  ];
+  store.updateStep.mockImplementation(async (_taskId: string, stepIndex: number, status: string) => {
+    const current = stepStates[stepIndex];
+    const isRegression = status === "in-progress" && (current.status === "done" || current.status === "skipped");
+    if (!isRegression) {
+      current.status = status;
+    }
+    return { steps: stepStates.map((s) => ({ ...s })) };
   });
   mockedExistsSync.mockReturnValue(true);
 
@@ -4675,10 +4684,13 @@ describe("Code review verdict enforcement - fn_task_update blocking", () => {
     mockedReviewStep.mockResolvedValue({ verdict: "REVISE", review: "Fix", summary: "Bad" });
 
     const tools = await captureTools();
-    await tools.fn_review_step("c1", { step: 1, type: "code", step_name: "Step1", baseline: "a" });
+    // Target step 3 (Testing, currently pending) so the in-progress transition is
+    // a valid forward move — the assertion below only verifies that a REVISE on
+    // the same step does not produce the "Cannot mark … as done" block.
+    await tools.fn_review_step("c1", { step: 3, type: "code", step_name: "Testing", baseline: "a" });
 
-    // "in-progress" should still work even with REVISE
-    const result = await tools.fn_task_update("c2", { step: 1, status: "in-progress" });
+    const result = await tools.fn_task_update("c2", { step: 3, status: "in-progress" });
+    expect(result.content[0].text).not.toContain("Cannot mark");
     expect(result.content[0].text).toContain("→ in-progress");
   });
 });