feat(HAI-114): remove --engine CLI flag and make engine always-on

- Remove --engine flag from dashboard command CLI options
- Make AI engine start automatically with the dashboard
- Update dashboard tests to reflect always-on engine behavior
- Update README.md to remove --engine references and examples

README.md

@@ -29,12 +29,9 @@ Tasks with dependencies are processed sequentially. Independent tasks run in par
 # Install dependencies
 pnpm install
 
-# Start the board (UI only)
+# Start the board (with AI engine)
 pnpm dev dashboard
 
-# Start with AI engine (auto-specify + auto-execute)
-pnpm dev dashboard -- --engine
-
 # Create a task via CLI
 pnpm dev task create "Fix the login redirect bug"
 
@@ -121,9 +118,9 @@ Every API response includes standard rate limit headers:
 
 When a client exceeds the limit, the API returns `429 Too Many Requests`.
 
-### AI Engine (`--engine`)
+### AI Engine
 
-When enabled, three components run:
+The AI engine starts automatically with the dashboard. Three components run:
 
 - **TriageProcessor** — Watches triage column. Spawns a pi agent session that reads the project, understands context, and writes a full PROMPT.md specification. Moves task to todo.
 
@@ -141,8 +138,7 @@ Each pi agent session gets:
 
 ```bash
 pnpm install
-pnpm dev dashboard              # Board only
-pnpm dev dashboard -- --engine  # Board + AI engine
+pnpm dev dashboard              # Board + AI engine
 pnpm dev task list              # CLI commands
 ```
 

packages/cli/src/bin.ts

@@ -59,7 +59,6 @@ Usage:
 
 Options:
   --port, -p <port>          Dashboard port (default: 4040)
-  --engine                   Enable AI engine (auto-specify + execute tasks)
   --attach <file>            Attach file(s) on task create (repeatable)
   --help, -h                 Show this help
 
@@ -87,9 +86,8 @@ async function main() {
         const portIdxShort = args.indexOf("-p");
         const pi = portIdx !== -1 ? portIdx : portIdxShort;
         const port = pi !== -1 ? parseInt(args[pi + 1], 10) : 4040;
-        const engine = args.includes("--engine");
         const open = !args.includes("--no-open");
-        await runDashboard(port, { engine, open });
+        await runDashboard(port, { open });
         break;
       }
 

packages/cli/src/commands/dashboard.test.ts

@@ -97,7 +97,7 @@ describe("runDashboard — WorktreePool wiring", () => {
   });
 
   it("passes a WorktreePool instance to TaskExecutor", async () => {
-    await runDashboard(0, { engine: true, open: false });
+    await runDashboard(0, { open: false });
 
     expect(capturedExecutorOpts).toBeDefined();
     expect(capturedExecutorOpts!.pool).toBeInstanceOf(WorktreePool);
@@ -107,7 +107,7 @@ describe("runDashboard — WorktreePool wiring", () => {
     const { aiMergeTask } = await import("@hai/engine");
     const { createServer } = await import("@hai/dashboard");
 
-    await runDashboard(0, { engine: false, open: false });
+    await runDashboard(0, { open: false });
 
     // rawMerge is exposed as the onMerge callback wired into createServer.
     const createServerCall = (createServer as ReturnType<typeof vi.fn>).mock.calls[0];
@@ -125,7 +125,7 @@ describe("runDashboard — WorktreePool wiring", () => {
     const { aiMergeTask } = await import("@hai/engine");
     const { createServer } = await import("@hai/dashboard");
 
-    await runDashboard(0, { engine: true, open: false });
+    await runDashboard(0, { open: false });
 
     // Trigger merger via onMerge
     const createServerCall = (createServer as ReturnType<typeof vi.fn>).mock.calls[0];
@@ -170,7 +170,7 @@ describe("runDashboard — auto-merge pause exclusion", () => {
       pollIntervalMs: 60_000,
     });
 
-    await runDashboard(0, { engine: true, open: false });
+    await runDashboard(0, { open: false });
 
     const { aiMergeTask } = await import("@hai/engine");
 
@@ -205,7 +205,7 @@ describe("runDashboard — auto-merge pause exclusion", () => {
       Promise.resolve({ merged: true }),
     );
 
-    await runDashboard(0, { engine: true, open: false });
+    await runDashboard(0, { open: false });
 
     // Give async handlers time to process
     await new Promise((r) => setTimeout(r, 50));

packages/cli/src/commands/dashboard.ts

@@ -12,7 +12,7 @@ function openBrowser(url: string): void {
   exec(cmd, () => {});
 }
 
-export async function runDashboard(port: number, opts: { engine?: boolean; open?: boolean } = {}) {
+export async function runDashboard(port: number, opts: { open?: boolean } = {}) {
   const cwd = process.cwd();
   const store = new TaskStore(cwd);
   await store.init();
@@ -143,16 +143,8 @@ export async function runDashboard(port: number, opts: { engine?: boolean; open?
   // Start the web server with AI merge, auth, and model registry wired in
   const app = createServer(store, { onMerge, authStorage, modelRegistry });
 
-  // Clean shutdown for file watcher when engine is not active
-  if (!opts.engine) {
-    process.on("SIGINT", () => {
-      store.stopWatching();
-      process.exit(0);
-    });
-  }
-
-  // Optionally start the AI engine
-  if (opts.engine) {
+  // Start the AI engine
+  {
     const triage = new TriageProcessor(store, cwd, {
       semaphore,
       onSpecifyStart: (t) => console.log(`[engine] Specifying ${t.id}...`),
@@ -242,13 +234,9 @@ export async function runDashboard(port: number, opts: { engine?: boolean; open?
     console.log();
     console.log(`  Tasks stored in .hai/tasks/`);
     console.log(`  Merge:      AI-assisted (conflict resolution + commit messages)`);
-    if (opts.engine) {
-      console.log(`  AI engine:  ✓ active`);
-      console.log(`    • triage: auto-specifying tasks`);
-      console.log(`    • scheduler: dependency-aware execution`);
-    } else {
-      console.log(`  AI engine:  off (use --engine to enable)`);
-    }
+    console.log(`  AI engine:  ✓ active`);
+    console.log(`    • triage: auto-specifying tasks`);
+    console.log(`    • scheduler: dependency-aware execution`);
     console.log(`  File watcher: ✓ active`);
     console.log(`  Press Ctrl+C to stop`);
     console.log();