From e23e1793f1013c7796f7456ddf9b78e3c0deffb9 Mon Sep 17 00:00:00 2001
From: Huaaudio <jiahua.zhang@kuleuven.be>
Date: Sat, 21 Mar 2026 07:21:33 +0100
Subject: [PATCH] remove leave tool due to delay

---
 pkg/agent/loop.go               |  3 --
 pkg/bus/types.go                |  3 +-
 pkg/channels/discord/discord.go | 11 +------
 pkg/tools/voice_leave.go        | 52 ---------------------------------
 pkg/voice/agent.go              | 19 +++++++++++-
 5 files changed, 20 insertions(+), 68 deletions(-)
 delete mode 100644 pkg/tools/voice_leave.go

diff --git a/pkg/agent/loop.go b/pkg/agent/loop.go
index aa15a23d5..725d42614 100644
--- a/pkg/agent/loop.go
+++ b/pkg/agent/loop.go
@@ -238,9 +238,6 @@ func registerSharedTools(
 			agent.Tools.Register(messageTool)
 		}
 
-		// Always register Voice Leave Tool (it inherently checks if channel == "discord")
-		agent.Tools.Register(tools.NewVoiceLeaveTool(msgBus))
-
 		// Send file tool (outbound media via MediaStore — store injected later by SetMediaStore)
 		if cfg.Tools.IsToolEnabled("send_file") {
 			sendFileTool := tools.NewSendFileTool(
diff --git a/pkg/bus/types.go b/pkg/bus/types.go
index b4e99d955..9c637f3e7 100644
--- a/pkg/bus/types.go
+++ b/pkg/bus/types.go
@@ -68,7 +68,6 @@ type AudioChunk struct {
 // VoiceControl represents state or commands for voice sessions.
 type VoiceControl struct {
 	SessionID string `json:"session_id"`
-	ChatID    string `json:"chat_id"`
-	Type      string `json:"type"`   // "state", "command"
+	Type      string `json:"type"`  // "state", "command"
 	Action    string `json:"action"` // "idle", "listening", "start", "stop"
 }
diff --git a/pkg/channels/discord/discord.go b/pkg/channels/discord/discord.go
index 6194685e8..26332455a 100644
--- a/pkg/channels/discord/discord.go
+++ b/pkg/channels/discord/discord.go
@@ -629,17 +629,8 @@ func (c *DiscordChannel) listenVoiceControl(ctx context.Context) {
 			return
 		case ctrl := <-c.bus.VoiceControlsChan():
 			if ctrl.Type == "command" && ctrl.Action == "leave" {
-				var guildID string
 				if strings.HasPrefix(ctrl.SessionID, "discord_vc_") {
-					guildID = strings.TrimPrefix(ctrl.SessionID, "discord_vc_")
-				} else if ctrl.ChatID != "" {
-					ch, err := c.session.State.Channel(ctrl.ChatID)
-					if err == nil {
-						guildID = ch.GuildID
-					}
-				}
-
-				if guildID != "" {
+					guildID := strings.TrimPrefix(ctrl.SessionID, "discord_vc_")
 					vc, exists := c.session.VoiceConnections[guildID]
 					if exists && vc != nil {
 						vc.Disconnect(ctx)
diff --git a/pkg/tools/voice_leave.go b/pkg/tools/voice_leave.go
deleted file mode 100644
index 882e7cb9d..000000000
--- a/pkg/tools/voice_leave.go
+++ /dev/null
@@ -1,52 +0,0 @@
-package tools
-
-import (
-	"context"
-
-	"github.com/sipeed/picoclaw/pkg/bus"
-	"github.com/sipeed/picoclaw/pkg/logger"
-)
-
-type VoiceLeaveTool struct {
-	bus *bus.MessageBus
-}
-
-func NewVoiceLeaveTool(mb *bus.MessageBus) *VoiceLeaveTool {
-	return &VoiceLeaveTool{bus: mb}
-}
-
-func (t *VoiceLeaveTool) Name() string {
-	return "voice_leave"
-}
-
-func (t *VoiceLeaveTool) Description() string {
-	return "Disconnects the bot from the current voice channel. Use this tool when the user says goodbye or explicitly asks you to leave the voice chat."
-}
-
-func (t *VoiceLeaveTool) Parameters() map[string]any {
-	return map[string]any{
-		"type":       "object",
-		"properties": map[string]any{},
-	}
-}
-
-func (t *VoiceLeaveTool) Execute(ctx context.Context, args map[string]any) *ToolResult {
-	channel := ToolChannel(ctx)
-	chatID := ToolChatID(ctx)
-
-	if channel != "discord" {
-		return &ToolResult{ForLLM: "Can only leave voice channels on Discord", IsError: true}
-	}
-
-	t.bus.PublishVoiceControl(ctx, bus.VoiceControl{
-		ChatID: chatID,
-		Type:   "command",
-		Action: "leave",
-	})
-
-	logger.InfoCF("agent", "Voice command triggered via tool: leave", map[string]any{"chat_id": chatID})
-
-	return &ToolResult{
-		ForLLM: "Successfully sent disconnect command to voice adapter.",
-	}
-}
diff --git a/pkg/voice/agent.go b/pkg/voice/agent.go
index 233da6019..6a2abd24f 100644
--- a/pkg/voice/agent.go
+++ b/pkg/voice/agent.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
+	"strings"
 	"sync"
 	"time"
 
@@ -184,7 +185,23 @@ func (a *Agent) processUtterance(ctx context.Context, acc *speechAccumulator) {
 
 	channelType := "discord"
 
-	oralPrompt := "\n\n[SYSTEM]: The user just spoke this to you over voice chat. Please reply in a highly concise, conversational, oral style suitable for text-to-speech. Do not use markdown, emojis, asterisks, or code blocks. Speak naturally. If the user expresses that they want to end the call, say goodbye and use the voice_leave tool."
+	text := strings.ToLower(strings.TrimSpace(res.Text))
+	if strings.Contains(text, "leave the voice channel") || strings.Contains(text, "leave voice") || strings.Contains(text, "disconnect voice") {
+		logger.InfoCF("voice-agent", "Voice command triggered: leave", nil)
+		a.bus.PublishVoiceControl(ctx, bus.VoiceControl{
+			SessionID: acc.sessionID,
+			Type:      "command",
+			Action:    "leave",
+		})
+		a.bus.PublishOutbound(ctx, bus.OutboundMessage{
+			Channel: channelType,
+			ChatID:  acc.chatID,
+			Content: "Goodbye! Leaving the voice channel.",
+		})
+		return
+	}
+
+	oralPrompt := "\n\n[SYSTEM]: The user just spoke this to you over voice chat. Please reply in a highly concise, conversational, oral style suitable for text-to-speech. Do not use markdown, emojis, asterisks, or code blocks. Speak naturally."
 
 	a.bus.PublishInbound(ctx, bus.InboundMessage{
 		Channel:  channelType,