fix: clean up activePerAgent before sending to resultCh

Move activePerAgent decrement/deletion out of execute() and
executeResume() defers and into the code paths immediately before each
resultCh send (handleRunResult and early-return paths). This guarantees
that when a result consumer reads from the channel the map is already
clean, eliminating a race between defer and result receipt.

Remove the polling loop from TestPool_ActivePerAgent_DeletesZeroEntries
and check the map state immediately after reading the result instead.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2 files changed, 27 insertions, 21 deletions

diff --git a/internal/executor/executor.go b/internal/executor/executor.go
index c07171b..219a40b 100644
--- a/internal/executor/executor.go
+++ b/internal/executor/executor.go
@@ -204,10 +204,6 @@ func (p *Pool) executeResume(ctx context.Context, t *task.Task, exec *storage.Ex
 	defer func() {
 		p.mu.Lock()
 		p.active--
-		p.activePerAgent[agentType]--
-		if p.activePerAgent[agentType] == 0 {
-			delete(p.activePerAgent, agentType)
-		}
 		p.mu.Unlock()
 		select {
 		case p.doneCh <- struct{}{}:
@@ -218,6 +214,12 @@ func (p *Pool) executeResume(ctx context.Context, t *task.Task, exec *storage.Ex
 	runner, err := p.getRunner(t)
 	if err != nil {
 		p.logger.Error("failed to get runner for resume", "error", err, "taskID", t.ID)
+		p.mu.Lock()
+		p.activePerAgent[agentType]--
+		if p.activePerAgent[agentType] == 0 {
+			delete(p.activePerAgent, agentType)
+		}
+		p.mu.Unlock()
 		p.resultCh <- &Result{TaskID: t.ID, Execution: exec, Err: err}
 		return
 	}
@@ -364,6 +366,12 @@ func (p *Pool) handleRunResult(ctx context.Context, t *task.Task, exec *storage.
 	if updateErr := p.store.UpdateExecution(exec); updateErr != nil {
 		p.logger.Error("failed to update execution", "error", updateErr)
 	}
+	p.mu.Lock()
+	p.activePerAgent[agentType]--
+	if p.activePerAgent[agentType] == 0 {
+		delete(p.activePerAgent, agentType)
+	}
+	p.mu.Unlock()
 	p.resultCh <- &Result{TaskID: t.ID, Execution: exec, Err: err}
 }
 
@@ -465,10 +473,6 @@ func (p *Pool) execute(ctx context.Context, t *task.Task) {
 	defer func() {
 		p.mu.Lock()
 		p.active--
-		p.activePerAgent[agentType]--
-		if p.activePerAgent[agentType] == 0 {
-			delete(p.activePerAgent, agentType)
-		}
 		p.mu.Unlock()
 		select {
 		case p.doneCh <- struct{}{}:
@@ -494,6 +498,12 @@ func (p *Pool) execute(ctx context.Context, t *task.Task) {
 		if err := p.store.UpdateTaskState(t.ID, task.StateFailed); err != nil {
 			p.logger.Error("failed to update task state", "taskID", t.ID, "state", task.StateFailed, "error", err)
 		}
+		p.mu.Lock()
+		p.activePerAgent[agentType]--
+		if p.activePerAgent[agentType] == 0 {
+			delete(p.activePerAgent, agentType)
+		}
+		p.mu.Unlock()
 		p.resultCh <- &Result{TaskID: t.ID, Execution: exec, Err: err}
 		return
 	}
@@ -516,6 +526,12 @@ func (p *Pool) execute(ctx context.Context, t *task.Task) {
 			if err := p.store.UpdateTaskState(t.ID, task.StateFailed); err != nil {
 				p.logger.Error("failed to update task state", "taskID", t.ID, "state", task.StateFailed, "error", err)
 			}
+			p.mu.Lock()
+			p.activePerAgent[agentType]--
+			if p.activePerAgent[agentType] == 0 {
+				delete(p.activePerAgent, agentType)
+			}
+			p.mu.Unlock()
 			p.resultCh <- &Result{TaskID: t.ID, Execution: exec, Err: err}
 			return
 		}
diff --git a/internal/executor/executor_test.go b/internal/executor/executor_test.go
index d8a2b77..878a32d 100644
--- a/internal/executor/executor_test.go
+++ b/internal/executor/executor_test.go
@@ -751,19 +751,9 @@ func TestPool_ActivePerAgent_DeletesZeroEntries(t *testing.T) {
 	pool.Submit(context.Background(), tk)
 	<-pool.Results()
 
-	// The deferred cleanup in execute() runs after resultCh is sent, so poll
-	// briefly for the map entry to be removed rather than checking immediately.
-	var exists bool
-	deadline := time.Now().Add(100 * time.Millisecond)
-	for time.Now().Before(deadline) {
-		pool.mu.Lock()
-		_, exists = pool.activePerAgent["claude"]
-		pool.mu.Unlock()
-		if !exists {
-			break
-		}
-		time.Sleep(time.Millisecond)
-	}
+	pool.mu.Lock()
+	_, exists := pool.activePerAgent["claude"]
+	pool.mu.Unlock()
 
 	if exists {
 		t.Error("activePerAgent should not have a zero-count entry for claude after task completes")