fix(loop-scheduler): await waitForIdle in agent_end before draining

Inside an agent_end listener, agent.state.isStreaming is still true —
finishRun() only clears it in the finally block of runWithLifecycle,
after all agent_end listeners settle. So when we dispatched a pending
job from agent_end and called pi.sendUserMessage(..., { deliverAs:
'followUp' }), the message was routed into agent.followUpQueue. The
agent loop had already passed its getFollowUpMessages() check, so it
exited without draining the queue. The message sat there as a stuck
'Follow-up: ...' in pi's UI, agentBusy stayed true forever, and every
subsequent pending loop was blocked because no further agent_end fired.

Await ctx.waitForIdle() in the agent_end handler before resetting
agentBusy and calling drainPendingJobs. By then finishRun() has cleared
isStreaming, so sendUserMessage starts a fresh run instead of enqueueing
into a dead followUp queue, and pending loops drain serially as designed.

Amp-Thread-ID: https://ampcode.com/threads/T-019e5de9-a0c3-7559-9cf0-f81ce751e763
Co-authored-by: Amp <amp@ampcode.com>

1 files changed, 12 insertions, 1 deletions

diff --git a/pi/agent/extensions/loop-scheduler/index.ts b/pi/agent/extensions/loop-scheduler/index.ts
index cb5398b..67247b8 100644
--- a/pi/agent/extensions/loop-scheduler/index.ts
+++ b/pi/agent/extensions/loop-scheduler/index.ts
@@ -1196,11 +1196,22 @@ export default function loopSchedulerExtension(pi: ExtensionAPI): void {
 
 	pi.on("agent_end", async (_event, ctx) => {
 		rememberContext(ctx);
-		agentBusy = false;
 		currentAssistantText = "";
 		currentIdleGeneration += 1;
 		queueIdleWatchJobs();
 		updateUi(ctx);
+		// CRITICAL: inside agent_end the agent's isStreaming flag is STILL true
+		// (finishRun() runs in the finally block after all listeners settle, see
+		// pi-coding-agent/packages/agent/src/agent.ts). If we dispatch here and
+		// call pi.sendUserMessage(..., { deliverAs: "followUp" }) right now, the
+		// message gets routed into agent.followUpQueue. But the agent loop has
+		// already passed its getFollowUpMessages() check — it will exit without
+		// draining the queue and our message sits there forever, visible as a
+		// stuck "Follow-up: ..." in pi's UI. We'd also leak agentBusy=true and
+		// block every subsequent pending job because no further agent_end fires.
+		// Wait for the run to actually finish before draining.
+		await ctx.waitForIdle();
+		agentBusy = false;
 		drainPendingJobs();
 	});