diff --git a/agent/protocol/__init__.py b/agent/protocol/__init__.py
index a9fe5a3e..f0a7a4e2 100644
--- a/agent/protocol/__init__.py
+++ b/agent/protocol/__init__.py
@@ -3,6 +3,11 @@ from .agent_stream import AgentStreamExecutor
 from .task import Task, TaskType, TaskStatus
 from .result import AgentResult, AgentAction, AgentActionType, ToolResult
 from .models import LLMModel, LLMRequest, ModelFactory
+from .cancel import (
+    AgentCancelledError,
+    CancelTokenRegistry,
+    get_cancel_registry,
+)
 
 __all__ = [
     'Agent', 
@@ -16,5 +21,8 @@ __all__ = [
     'ToolResult',
     'LLMModel',
     'LLMRequest', 
-    'ModelFactory'
-]
\ No newline at end of file
+    'ModelFactory',
+    'AgentCancelledError',
+    'CancelTokenRegistry',
+    'get_cancel_registry',
+]
diff --git a/agent/protocol/agent.py b/agent/protocol/agent.py
index 285a9732..d944660b 100644
--- a/agent/protocol/agent.py
+++ b/agent/protocol/agent.py
@@ -365,7 +365,8 @@ class Agent:
 
         return action
 
-    def run_stream(self, user_message: str, on_event=None, clear_history: bool = False, skill_filter=None) -> str:
+    def run_stream(self, user_message: str, on_event=None, clear_history: bool = False,
+                   skill_filter=None, cancel_event=None) -> str:
         """
         Execute single agent task with streaming (based on tool-call)
 
@@ -374,6 +375,7 @@ class Agent:
         - Multi-turn reasoning based on tool-call
         - Event callbacks
         - Persistent conversation history across calls
+        - User-initiated cancellation via ``cancel_event``
 
         Args:
             user_message: User message
@@ -381,6 +383,11 @@ class Agent:
                      event = {"type": str, "timestamp": float, "data": dict}
             clear_history: If True, clear conversation history before this call (default: False)
             skill_filter: Optional list of skill names to include in this run
+            cancel_event: Optional threading.Event polled at agent checkpoints.
+                When set, the loop exits at the next safe point, injects a
+                "[Interrupted by user]" assistant note, and returns the
+                partial response. ``messages`` stays in a valid state
+                (tool_use/tool_result pairs preserved).
 
         Returns:
             Final response text
@@ -424,7 +431,8 @@ class Agent:
             max_turns=self.max_steps,
             on_event=on_event,
             messages=messages_copy,  # Pass copied message history
-            max_context_turns=max_context_turns
+            max_context_turns=max_context_turns,
+            cancel_event=cancel_event,
         )
 
         # Execute
diff --git a/agent/protocol/agent_stream.py b/agent/protocol/agent_stream.py
index ef4f975b..701b00de 100644
--- a/agent/protocol/agent_stream.py
+++ b/agent/protocol/agent_stream.py
@@ -7,6 +7,7 @@ import json
 import time
 from typing import List, Dict, Any, Optional, Callable, Tuple
 
+from agent.protocol.cancel import AgentCancelledError
 from agent.protocol.models import LLMRequest, LLMModel
 from agent.protocol.message_utils import sanitize_claude_messages, compress_turn_to_text_only
 from agent.tools.base_tool import BaseTool, ToolResult
@@ -64,7 +65,8 @@ class AgentStreamExecutor:
             max_turns: int = 50,
             on_event: Optional[Callable] = None,
             messages: Optional[List[Dict]] = None,
-            max_context_turns: int = 30
+            max_context_turns: int = 30,
+            cancel_event=None,
     ):
         """
         Initialize stream executor
@@ -78,6 +80,10 @@ class AgentStreamExecutor:
             on_event: Event callback function
             messages: Optional existing message history (for persistent conversations)
             max_context_turns: Maximum number of conversation turns to keep in context
+            cancel_event: Optional threading.Event used to signal user cancel.
+                Checked at every safe point (turn boundary, before tool execution,
+                during LLM streaming). When set, raises AgentCancelledError which
+                run_stream catches to gracefully wind down.
         """
         self.agent = agent
         self.model = model
@@ -87,6 +93,7 @@ class AgentStreamExecutor:
         self.max_turns = max_turns
         self.on_event = on_event
         self.max_context_turns = max_context_turns
+        self.cancel_event = cancel_event
 
         # Message history - use provided messages or create new list
         self.messages = messages if messages is not None else []
@@ -97,6 +104,73 @@ class AgentStreamExecutor:
         # Track files to send (populated by read tool)
         self.files_to_send = []  # List of file metadata dicts
 
+    def _check_cancelled(self) -> None:
+        """Raise AgentCancelledError if the user requested cancellation.
+
+        Called at safe points (turn start, between tool calls, between LLM
+        chunks). Cheap to call: just an Event.is_set() probe.
+        """
+        if self.cancel_event is not None and self.cancel_event.is_set():
+            raise AgentCancelledError("agent cancelled by user")
+
+    def _handle_cancelled(self, partial_response: str) -> None:
+        """Wind down ``self.messages`` after a user-initiated cancel.
+
+        The messages list may be in any of these states when we get here:
+          (a) Last message is an assistant message containing tool_use
+              blocks but the matching tool_result has not been appended yet.
+          (b) Last message is an assistant text-only reply (cancel happened
+              right before the next turn started).
+          (c) Last message is a user tool_result message and we cancelled
+              between turns.
+
+        For (a) we MUST synthesise tool_result blocks, otherwise the next
+        request will fail Claude/OpenAI's strict pairing validation. For
+        (b)/(c) the state is already valid and we just append a small
+        cancellation note so the user/LLM both see the boundary clearly.
+        """
+        try:
+            # Step 1: close any orphaned tool_use in the trailing assistant
+            # message by injecting matching tool_result blocks.
+            if self.messages and isinstance(self.messages[-1], dict) \
+                    and self.messages[-1].get("role") == "assistant":
+                last = self.messages[-1]
+                content = last.get("content")
+                if isinstance(content, list):
+                    pending_tool_use_ids = [
+                        block.get("id")
+                        for block in content
+                        if isinstance(block, dict) and block.get("type") == "tool_use"
+                    ]
+                    pending_tool_use_ids = [tid for tid in pending_tool_use_ids if tid]
+                    if pending_tool_use_ids:
+                        tool_result_blocks = [
+                            {
+                                "type": "tool_result",
+                                "tool_use_id": tid,
+                                "content": "Cancelled by user before this tool finished.",
+                                "is_error": True,
+                            }
+                            for tid in pending_tool_use_ids
+                        ]
+                        self.messages.append({
+                            "role": "user",
+                            "content": tool_result_blocks,
+                        })
+                        logger.info(
+                            f"[Agent] Injected {len(tool_result_blocks)} cancellation "
+                            f"tool_result blocks to keep message history valid"
+                        )
+
+            # Step 2: append a stable "interrupted" marker so the LLM sees a
+            # clear stop boundary on the next turn.
+            self.messages.append({
+                "role": "assistant",
+                "content": [{"type": "text", "text": "_(Cancelled by user)_"}],
+            })
+        except Exception as e:
+            logger.warning(f"[Agent] _handle_cancelled cleanup failed: {e}")
+
     def _emit_event(self, event_type: str, data: dict = None):
         """Emit event"""
         if self.on_event:
@@ -270,8 +344,13 @@ class AgentStreamExecutor:
         final_response = ""
         turn = 0
 
+        cancelled = False
         try:
             while turn < self.max_turns:
+                # Check at the very top of every turn so a cancel arriving
+                # between turns short-circuits cleanly.
+                self._check_cancelled()
+
                 turn += 1
                 logger.info(f"[Agent] 第 {turn} 轮")
                 self._emit_event("turn_start", {"turn": turn})
@@ -375,6 +454,8 @@ class AgentStreamExecutor:
 
                 try:
                     for tool_call in tool_calls:
+                        # Honour cancel between tool invocations within the same turn
+                        self._check_cancelled()
                         result = self._execute_tool(tool_call)
                         tool_results.append(result)
                         
@@ -557,6 +638,15 @@ class AgentStreamExecutor:
                         self.messages.pop(prompt_insert_idx)
                         logger.debug("[Agent] Removed injected max-steps prompt from message history")
 
+        except AgentCancelledError:
+            # User-initiated stop: wind down message history cleanly so the
+            # next turn is unaffected; channels emit a "cancelled" UI event.
+            cancelled = True
+            logger.info(f"[Agent] 🛑 已被用户中止 (第 {turn} 轮)")
+            self._handle_cancelled(final_response)
+            if not final_response or not final_response.strip():
+                final_response = "_(Cancelled)_"
+
         except Exception as e:
             logger.error(f"❌ Agent执行错误: {e}")
             self._emit_event("error", {"error": str(e)})
@@ -564,8 +654,11 @@ class AgentStreamExecutor:
 
         finally:
             final_response = final_response.strip() if final_response else final_response
-            logger.info(f"[Agent] 🏁 完成 ({turn}轮)")
-            self._emit_event("agent_end", {"final_response": final_response})
+            if cancelled:
+                # Emit before agent_end so channels can mark UI as cancelled
+                self._emit_event("agent_cancelled", {"final_response": final_response})
+            logger.info(f"[Agent] 🏁 完成 ({turn}轮)" + (" [cancelled]" if cancelled else ""))
+            self._emit_event("agent_end", {"final_response": final_response, "cancelled": cancelled})
 
         return final_response
 
@@ -644,7 +737,32 @@ class AgentStreamExecutor:
         try:
             stream = self.model.call_stream(request)
 
+            # Probe cancel every N chunks to bound reaction time without
+            # checking on every token.
+            _cancel_probe_counter = 0
+            _CANCEL_PROBE_EVERY = 8
+
             for chunk in stream:
+                _cancel_probe_counter += 1
+                if _cancel_probe_counter >= _CANCEL_PROBE_EVERY:
+                    _cancel_probe_counter = 0
+                    if self.cancel_event is not None and self.cancel_event.is_set():
+                        # Persist partial text only; tool_use args may be
+                        # truncated mid-stream and would fail validation.
+                        logger.info("[Agent] cancel detected mid-stream, aborting LLM call")
+                        if full_content:
+                            partial_msg = {
+                                "role": "assistant",
+                                "content": [{"type": "text", "text": full_content}],
+                            }
+                            self.messages.append(partial_msg)
+                        self._emit_event("message_end", {
+                            "content": full_content,
+                            "tool_calls": [],
+                            "cancelled": True,
+                        })
+                        raise AgentCancelledError("cancelled during LLM streaming")
+
                 # Check for errors
                 if isinstance(chunk, dict) and chunk.get("error"):
                     # Extract error message from nested structure
@@ -738,6 +856,10 @@ class AgentStreamExecutor:
                     elif isinstance(choice, dict) and choice.get("_gemini_raw_parts"):
                         gemini_raw_parts = choice["_gemini_raw_parts"]
 
+        except AgentCancelledError:
+            # Must propagate untouched; never treat as a retryable error.
+            raise
+
         except Exception as e:
             error_str = str(e)
             error_str_lower = error_str.lower()
diff --git a/agent/protocol/cancel.py b/agent/protocol/cancel.py
new file mode 100644
index 00000000..6354cd38
--- /dev/null
+++ b/agent/protocol/cancel.py
@@ -0,0 +1,121 @@
+"""
+Cancel token registry for aborting in-flight agent runs.
+
+A user cancel (web Cancel button, /cancel command) sets a threading.Event
+that the agent loop polls at safe checkpoints. Tokens are keyed by
+request_id (preferred) and tracked under session_id as a fallback. Entries
+are released after the run completes to keep the registry bounded.
+
+No project deps — importable from any layer without circular imports.
+"""
+
+from __future__ import annotations
+
+import threading
+from typing import Dict, Optional
+
+
+class AgentCancelledError(Exception):
+    """Raised inside the agent loop when a stop has been requested.
+
+    The agent stream executor catches this, injects a "[Interrupted]" note
+    into the message history (preserving tool_use/tool_result integrity)
+    and returns a partial response to the caller.
+    """
+
+
+class _CancelEntry:
+    __slots__ = ("event", "session_id")
+
+    def __init__(self, session_id: Optional[str]):
+        self.event = threading.Event()
+        self.session_id = session_id
+
+
+class CancelTokenRegistry:
+    """In-process registry mapping request_id -> cancel Event.
+
+    Thread-safe. Singleton via module-level ``_registry``.
+    """
+
+    def __init__(self):
+        self._lock = threading.Lock()
+        self._by_request: Dict[str, _CancelEntry] = {}
+        # session_id -> set of request_ids currently in flight (usually 1).
+        self._by_session: Dict[str, set] = {}
+
+    def register(self, request_id: str, session_id: Optional[str] = None) -> threading.Event:
+        """Create (or return existing) cancel event for a request.
+
+        Returns the threading.Event the caller should poll via ``is_set()``.
+        """
+        if not request_id:
+            return threading.Event()
+        with self._lock:
+            entry = self._by_request.get(request_id)
+            if entry is None:
+                entry = _CancelEntry(session_id)
+                self._by_request[request_id] = entry
+                if session_id:
+                    self._by_session.setdefault(session_id, set()).add(request_id)
+            return entry.event
+
+    def get_event(self, request_id: str) -> Optional[threading.Event]:
+        if not request_id:
+            return None
+        with self._lock:
+            entry = self._by_request.get(request_id)
+            return entry.event if entry else None
+
+    def cancel_request(self, request_id: str) -> bool:
+        """Trigger cancel for a specific request. Returns True when matched."""
+        if not request_id:
+            return False
+        with self._lock:
+            entry = self._by_request.get(request_id)
+        if entry is None:
+            return False
+        entry.event.set()
+        return True
+
+    def cancel_session(self, session_id: str) -> int:
+        """Trigger cancel for every in-flight request of a session.
+
+        Returns the number of requests cancelled (0 when nothing was running).
+        """
+        if not session_id:
+            return 0
+        with self._lock:
+            request_ids = list(self._by_session.get(session_id, ()))
+            entries = [self._by_request[r] for r in request_ids if r in self._by_request]
+        for entry in entries:
+            entry.event.set()
+        return len(entries)
+
+    def unregister(self, request_id: str) -> None:
+        """Remove an entry once the agent run is done. Safe to call twice."""
+        if not request_id:
+            return
+        with self._lock:
+            entry = self._by_request.pop(request_id, None)
+            if entry and entry.session_id:
+                bucket = self._by_session.get(entry.session_id)
+                if bucket is not None:
+                    bucket.discard(request_id)
+                    if not bucket:
+                        self._by_session.pop(entry.session_id, None)
+
+    def has_active(self, session_id: str) -> bool:
+        if not session_id:
+            return False
+        with self._lock:
+            bucket = self._by_session.get(session_id)
+            return bool(bucket)
+
+
+_registry = CancelTokenRegistry()
+
+
+def get_cancel_registry() -> CancelTokenRegistry:
+    """Module-level accessor for the singleton registry."""
+    return _registry
diff --git a/bridge/agent_bridge.py b/bridge/agent_bridge.py
index a20659a8..a924dab2 100644
--- a/bridge/agent_bridge.py
+++ b/bridge/agent_bridge.py
@@ -5,7 +5,7 @@ Agent Bridge - Integrates Agent system with existing COW bridge
 import os
 from typing import Optional, List
 
-from agent.protocol import Agent, LLMModel, LLMRequest
+from agent.protocol import Agent, LLMModel, LLMRequest, get_cancel_registry
 from bridge.agent_event_handler import AgentEventHandler
 from bridge.agent_initializer import AgentInitializer
 from bridge.bridge import Bridge
@@ -399,11 +399,22 @@ class AgentBridge:
         """
         session_id = None
         agent = None
+        request_id = None
+        cancel_event = None
         try:
             # Extract session_id from context for user isolation
             if context:
                 session_id = context.kwargs.get("session_id") or context.get("session_id")
-            
+                request_id = context.kwargs.get("request_id") or context.get("request_id")
+
+            # Register a cancel token. Prefer per-turn request_id (web),
+            # fall back to session_id (IM channels). The Event is polled by
+            # AgentStreamExecutor at safe checkpoints.
+            registry = get_cancel_registry()
+            token_key = request_id or session_id
+            if token_key:
+                cancel_event = registry.register(token_key, session_id=session_id)
+
             # Get agent for this session (will auto-initialize if needed)
             agent = self.get_agent(session_id=session_id)
             if not agent:
@@ -458,7 +469,8 @@ class AgentBridge:
                 response = agent.run_stream(
                     user_message=query,
                     on_event=event_handler.handle_event,
-                    clear_history=clear_history
+                    clear_history=clear_history,
+                    cancel_event=cancel_event,
                 )
             finally:
                 # Restore original tools
@@ -468,6 +480,13 @@ class AgentBridge:
                 # Log execution summary
                 event_handler.log_summary()
 
+                # Release cancel token; keep registry bounded.
+                if token_key:
+                    try:
+                        registry.unregister(token_key)
+                    except Exception:
+                        pass
+
             # Persist new messages generated during this run
             if session_id:
                 channel_type = (context.get("channel_type") or "") if context else ""
@@ -521,6 +540,12 @@ class AgentBridge:
                         logger.info(f"[AgentBridge] Cleared DB for session after error: {session_id}")
                 except Exception as db_err:
                     logger.warning(f"[AgentBridge] Failed to clear DB after error: {db_err}")
+            # Release cancel token on error path too (idempotent).
+            if cancel_event is not None and (request_id or session_id):
+                try:
+                    get_cancel_registry().unregister(request_id or session_id)
+                except Exception:
+                    pass
             return Reply(ReplyType.ERROR, f"Agent error: {str(e)}")
     
     def _schedule_mcp_hot_reload(self, agent):
diff --git a/channel/chat_channel.py b/channel/chat_channel.py
index c38dd7c8..6a9a1952 100644
--- a/channel/chat_channel.py
+++ b/channel/chat_channel.py
@@ -438,8 +438,21 @@ class ChatChannel(Channel):
 
         return func
 
+    # Chat commands that must bypass the per-session serial queue,
+    # otherwise /cancel would queue behind the task it tries to cancel.
+    # Use /cancel (not /stop) to avoid colliding with `cow stop` CLI.
+    _BYPASS_QUEUE_COMMANDS = ("/cancel",)
+
     def produce(self, context: Context):
         session_id = context["session_id"]
+
+        # Fast path: /cancel must not enter the queue.
+        if context.type == ContextType.TEXT and context.content:
+            stripped = context.content.strip().lower()
+            if stripped in self._BYPASS_QUEUE_COMMANDS:
+                self._handle_cancel_command(context, session_id)
+                return
+
         with self.lock:
             if session_id not in self.sessions:
                 self.sessions[session_id] = [
@@ -451,6 +464,29 @@ class ChatChannel(Channel):
             else:
                 self.sessions[session_id][0].put(context)
 
+    def _handle_cancel_command(self, context: Context, session_id: str) -> None:
+        """Cancel any in-flight agent run for *session_id* and reply inline.
+
+        Runs synchronously on the caller's thread. Reply is sent through
+        _send_reply so plugins (e.g. logging) still observe it.
+        """
+        try:
+            from agent.protocol import get_cancel_registry
+            from bridge.reply import Reply, ReplyType
+
+            cancelled = get_cancel_registry().cancel_session(session_id)
+            text = (
+                "🛑 已中止"
+                if cancelled > 0
+                else "当前没有可中止的任务。"
+            )
+            logger.info(
+                f"[chat_channel] /cancel fast-path: session={session_id}, cancelled={cancelled}"
+            )
+            self._send_reply(context, Reply(ReplyType.TEXT, text))
+        except Exception as e:
+            logger.warning(f"[chat_channel] /cancel fast-path failed: {e}")
+
     # 消费者函数，单独线程，用于从消息队列中取出消息并处理
     def consume(self):
         while True:
diff --git a/channel/feishu/feishu_channel.py b/channel/feishu/feishu_channel.py
index ca18e64b..9a9f3307 100644
--- a/channel/feishu/feishu_channel.py
+++ b/channel/feishu/feishu_channel.py
@@ -752,6 +752,9 @@ class FeiShuChanel(ChatChannel):
         init_in_flight = [False]
         # 一旦初始化失败就长期标记为 disabled，本次回复不再尝试任何流式调用
         disabled = [False]
+        # True after agent_cancelled: agent_end stops rewriting the card
+        # with stale final_response and just finalizes current content.
+        cancelled = [False]
         lock = threading.Lock()
 
         # ---- 异步推送队列 ----------------------------------------------------
@@ -1076,18 +1079,42 @@ class FeiShuChanel(ChatChannel):
                     message_id[0] = None
                     sequence[0] = 0
 
+            elif event_type == "agent_cancelled":
+                # Lock channel into "no-rewrite" mode: the subsequent
+                # agent_end's final_response is from the last *completed*
+                # turn (the user already saw it), so rewriting the card
+                # would duplicate it visually.
+                with lock:
+                    cancelled[0] = True
+
             elif event_type == "agent_end":
                 # 最终回复：用 final_response 覆盖当前流式卡片，然后关闭流式模式。
                 final_response = data.get("final_response", "")
-                if not final_response:
-                    return
-                final_text = str(final_response)
                 # 标记 streamed 让 chat_channel 跳过 send()
                 context["feishu_streamed"] = True
 
                 with lock:
+                    was_cancelled = cancelled[0]
                     has_card = card_id[0] is not None
                     init_busy = init_in_flight[0]
+                    pending_text = current_text[0]
+
+                if was_cancelled:
+                    # Cancelled path: finalize the in-flight card with
+                    # partial output (or a short marker if empty); drop
+                    # stale final_response to avoid duplicating last turn.
+                    if has_card:
+                        _drain_push_queue()
+                        partial = (pending_text or "").rstrip()
+                        final_text = partial or "_(已中止)_"
+                        _stream_update_text(final_text)
+                        _close_streaming_mode(final_text)
+                    push_queue.put(None)
+                    return
+
+                if not final_response:
+                    return
+                final_text = str(final_response)
 
                 # 罕见情况：agent_end 触发时还没创建过卡片（极快返回 / 没有
                 # message_update），主动创建一张承载 final_text。
diff --git a/channel/web/chat.html b/channel/web/chat.html
index 947e07b7..d90adb15 100644
--- a/channel/web/chat.html
+++ b/channel/web/chat.html
@@ -445,7 +445,7 @@
                                                bg-primary-400 text-white hover:bg-primary-500
                                                disabled:bg-slate-300 dark:disabled:bg-slate-600
                                                disabled:cursor-not-allowed cursor-pointer transition-colors duration-150"
-                                        disabled onclick="sendMessage()">
+                                        disabled>
                                     <i class="fas fa-paper-plane text-sm"></i>
                                 </button>
                             </div>
diff --git a/channel/web/static/css/console.css b/channel/web/static/css/console.css
index 35fc307b..957db3e0 100644
--- a/channel/web/static/css/console.css
+++ b/channel/web/static/css/console.css
@@ -1367,3 +1367,35 @@
     text-align: right;
 }
 .voice-pill audio { display: none; }
+
+/* Send button toggles into a Stop button while an SSE stream is in flight.
+   Match the look of the disabled send button (light grey block + white
+   glyph) so it reads as the same visual element, just paused/idle from
+   sending perspective and clickable to stop. */
+#send-btn.send-btn-cancel {
+    background-color: rgb(203 213 225) !important; /* slate-300, == disabled send-btn */
+    color: white !important;
+}
+#send-btn.send-btn-cancel:hover {
+    background-color: rgb(148 163 184) !important; /* slate-400 */
+}
+#send-btn.send-btn-cancel:disabled {
+    background-color: rgb(226 232 240) !important; /* slate-200, while stop is in flight */
+    color: white !important;
+    cursor: progress;
+}
+.dark #send-btn.send-btn-cancel {
+    background-color: rgb(71 85 105) !important; /* slate-600, == dark disabled send-btn */
+    color: white !important;
+}
+.dark #send-btn.send-btn-cancel:hover {
+    background-color: rgb(100 116 139) !important; /* slate-500 */
+}
+.dark #send-btn.send-btn-cancel:disabled {
+    background-color: rgb(51 65 85) !important; /* slate-700 */
+    color: rgb(203 213 225) !important;
+}
+
+.agent-cancelled-tag {
+    font-style: italic;
+}
diff --git a/channel/web/static/js/console.js b/channel/web/static/js/console.js
index 7376e29c..6d0a66fc 100644
--- a/channel/web/static/js/console.js
+++ b/channel/web/static/js/console.js
@@ -1016,7 +1016,60 @@ const inputHistory = [];
 let historyIdx = -1;
 let historySavedDraft = '';
 
+// While an SSE stream is in flight, the send button morphs into a cancel
+// button. Only one in-flight request is supported at a time.
+let activeRequestId = null;
+let sendBtnMode = 'send'; // 'send' | 'cancel'
+
+function setSendBtnCancelMode(requestId) {
+    activeRequestId = requestId;
+    sendBtnMode = 'cancel';
+    sendBtn.disabled = false;
+    sendBtn.classList.add('send-btn-cancel');
+    sendBtn.title = (currentLang === 'zh' ? '中止' : 'Cancel');
+    sendBtn.innerHTML = '<i class="fas fa-stop text-sm"></i>';
+}
+
+function resetSendBtnSendMode() {
+    activeRequestId = null;
+    sendBtnMode = 'send';
+    sendBtn.classList.remove('send-btn-cancel');
+    sendBtn.title = '';
+    sendBtn.innerHTML = '<i class="fas fa-paper-plane text-sm"></i>';
+    updateSendBtnState();
+}
+
+function requestCancel() {
+    const reqId = activeRequestId;
+    if (!reqId) return;
+    fetch('/cancel', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({ request_id: reqId, session_id: sessionId, lang: currentLang }),
+    }).catch(err => {
+        console.warn('[cancel] request failed', err);
+    });
+    // Optimistic UI lock so the click visibly registers before the SSE
+    // "cancelled" event arrives.
+    sendBtn.disabled = true;
+    sendBtn.title = (currentLang === 'zh' ? '已中止' : 'Cancelled');
+}
+
+// Button click is the only path to Cancel. Pressing Enter still calls
+// sendMessage() so users can submit "/cancel" as a regular slash command.
+sendBtn.addEventListener('click', () => {
+    if (sendBtnMode === 'cancel') {
+        requestCancel();
+    } else {
+        sendMessage();
+    }
+});
+
 function updateSendBtnState() {
+    if (sendBtnMode === 'cancel') {
+        // Don't downgrade a Cancel button on input edits.
+        return;
+    }
     sendBtn.disabled = uploadingCount > 0 || (!chatInput.value.trim() && pendingAttachments.length === 0);
 }
 
@@ -1264,6 +1317,7 @@ const SLASH_COMMANDS = [
     { cmd: '/knowledge on',        desc: '开启知识库' },
     { cmd: '/knowledge off',       desc: '关闭知识库' },
     { cmd: '/config',              desc: '查看当前配置' },
+    { cmd: '/cancel',              desc: '中止当前正在运行的 Agent 任务' },
     { cmd: '/logs',                desc: '查看最近日志' },
     { cmd: '/version',             desc: '查看版本' },
 ];
@@ -1534,6 +1588,7 @@ function sendVoiceMessage(text, audioUrl) {
         stream: true,
         timestamp: timestamp.toISOString(),
         is_voice: true,
+        lang: currentLang,
     };
 
     const MAX_RETRIES = 2;
@@ -1547,7 +1602,12 @@ function sendVoiceMessage(text, audioUrl) {
         .then(r => r.json())
         .then(data => {
             if (data.status === 'success') {
-                if (data.stream) {
+                if (data.inline_reply) {
+                    // Synchronous fast-path reply (e.g. /cancel); skip SSE.
+                    loadingEl.remove();
+                    addBotMessage(data.inline_reply, new Date());
+                } else if (data.stream) {
+                    setSendBtnCancelMode(data.request_id);
                     startSSE(data.request_id, loadingEl, timestamp, titleInfo);
                 } else {
                     loadingContainers[data.request_id] = loadingEl;
@@ -1555,6 +1615,7 @@ function sendVoiceMessage(text, audioUrl) {
             } else {
                 loadingEl.remove();
                 addBotMessage(t('error_send'), new Date());
+                resetSendBtnSendMode();
             }
         })
         .catch(err => {
@@ -1591,6 +1652,10 @@ function addUserVoiceMessage(audioUrl, caption, timestamp) {
 }
 
 function sendMessage() {
+    // Do NOT branch on sendBtnMode here: Enter should always send (so
+    // typing "/cancel" submits normally). Cancel is wired only to the
+    // send button's pointer click — see send-btn listener above.
+
     const text = chatInput.value.trim();
     if (!text && pendingAttachments.length === 0) return;
 
@@ -1619,7 +1684,7 @@ function sendMessage() {
     renderAttachmentPreview();
     sendBtn.disabled = true;
 
-    const body = { session_id: sessionId, message: text, stream: true, timestamp: timestamp.toISOString() };
+    const body = { session_id: sessionId, message: text, stream: true, timestamp: timestamp.toISOString(), lang: currentLang };
     if (attachments.length > 0) {
         body.attachments = attachments.map(a => ({
             file_path: a.file_path,
@@ -1641,7 +1706,13 @@ function sendMessage() {
         .then(r => r.json())
         .then(data => {
             if (data.status === 'success') {
-                if (data.stream) {
+                if (data.inline_reply) {
+                    // Channel handled synchronously (e.g. /cancel fast-path);
+                    // render as a bot bubble and skip SSE entirely.
+                    loadingEl.remove();
+                    addBotMessage(data.inline_reply, new Date());
+                } else if (data.stream) {
+                    setSendBtnCancelMode(data.request_id);
                     startSSE(data.request_id, loadingEl, timestamp, titleInfo);
                 } else {
                     loadingContainers[data.request_id] = loadingEl;
@@ -1649,12 +1720,14 @@ function sendMessage() {
             } else {
                 loadingEl.remove();
                 addBotMessage(t('error_send'), new Date());
+                resetSendBtnSendMode();
             }
         })
         .catch(err => {
             if (err.name === 'AbortError') {
                 loadingEl.remove();
                 addBotMessage(t('error_timeout'), new Date());
+                resetSendBtnSendMode();
                 return;
             }
             if (attempt < MAX_RETRIES) {
@@ -1664,6 +1737,7 @@ function sendMessage() {
             }
             loadingEl.remove();
             addBotMessage(t('error_send'), new Date());
+            resetSendBtnSendMode();
         });
     }
 
@@ -1919,14 +1993,33 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo) {
                 stepsEl.appendChild(wrap);
                 scrollChatToBottom();
 
+            } else if (item.type === 'cancelled') {
+                // Agent acknowledged the stop; mark the bubble. A trailing
+                // "done" still arrives with the partial answer.
+                ensureBotEl();
+                if (currentReasoningEl) {
+                    finalizeThinking(currentReasoningEl, reasoningStartTime, reasoningText);
+                    currentReasoningEl = null;
+                    reasoningText = '';
+                }
+                if (!botEl.querySelector('.agent-cancelled-tag')) {
+                    const tag = document.createElement('div');
+                    tag.className = 'agent-cancelled-tag text-xs text-amber-600 dark:text-amber-400 mt-1';
+                    tag.textContent = (currentLang === 'zh') ? '已中止' : 'Cancelled';
+                    stepsEl.appendChild(tag);
+                }
+                resetSendBtnSendMode();
+
             } else if (item.type === 'done') {
                 // Don't close the stream yet: the backend keeps it open
                 // for a short tail to deliver async attachments such as
                 // TTS audio (`voice_attach`). It will close the stream on
                 // its own via onerror once the tail expires.
                 done = true;
+                resetSendBtnSendMode();
 
-                const finalText = item.content || accumulatedText;
+                const finalTextRaw = item.content || accumulatedText;
+                const finalText = localizeCancelMarker(finalTextRaw);
 
                 if (!botEl && finalText) {
                     if (loadingEl) { loadingEl.remove(); loadingEl = null; }
@@ -1934,7 +2027,7 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo) {
                 } else if (botEl) {
                     contentEl.classList.remove('sse-streaming');
                     if (finalText) contentEl.innerHTML = renderMarkdown(finalText);
-                    contentEl.dataset.rawMd = finalText || '';
+                    contentEl.dataset.rawMd = finalTextRaw || '';
                     const copyBtn = botEl.querySelector('.copy-msg-btn');
                     if (copyBtn && finalText) copyBtn.style.display = '';
                     applyHighlighting(botEl);
@@ -1964,6 +2057,7 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo) {
                 delete activeStreams[requestId];
                 if (loadingEl) { loadingEl.remove(); loadingEl = null; }
                 addBotMessage(t('error_send'), new Date());
+                resetSendBtnSendMode();
             }
         };
 
@@ -2000,6 +2094,7 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo) {
                 applyHighlighting(botEl);
                 bindChatKnowledgeLinks(botEl);
             }
+            resetSendBtnSendMode();
         };
     }
 
@@ -2238,13 +2333,23 @@ function _renderSentFileFromToolResult(step) {
         `<i class="fas fa-file-download" style="color:#6b7280;"></i> ${escapeHtml(fileName)}</a></div>`;
 }
 
+// Cosmetic translator for cancel markers persisted in history.
+// History keeps the English canonical form for the LLM; only display is localized.
+function localizeCancelMarker(text) {
+    if (!text) return text;
+    if (currentLang !== 'zh') return text;
+    return text
+        .replace(/_\(Cancelled by user\)_/g, '_(用户已中止)_')
+        .replace(/_\(Cancelled\)_/g, '_(已中止)_');
+}
+
 function createBotMessageEl(content, timestamp, requestId, msg) {
     const el = document.createElement('div');
     el.className = 'flex gap-3 px-4 sm:px-6 py-3';
     if (requestId) el.dataset.requestId = requestId;
 
     let stepsHtml = '';
-    let displayContent = content;
+    let displayContent = localizeCancelMarker(content);
 
     if (msg && msg.steps && msg.steps.length > 0) {
         // New format: ordered steps with interleaved content
diff --git a/channel/web/web_channel.py b/channel/web/web_channel.py
index 371f0271..ab1d6915 100644
--- a/channel/web/web_channel.py
+++ b/channel/web/web_channel.py
@@ -93,6 +93,15 @@ def _require_auth():
                             json.dumps({"status": "error", "message": "Unauthorized"}))
 
 
+# Localized text for /cancel system replies. Web is the only channel that
+# honors a per-request `lang`; other channels reply in Chinese by default.
+def _cancel_reply_text(cancelled: int, lang: str) -> str:
+    en = lang.startswith("en")
+    if cancelled > 0:
+        return "🛑 Cancelled." if en else "🛑 已中止"
+    return "Nothing to cancel." if en else "当前没有可中止的任务。"
+
+
 def _get_upload_dir() -> str:
     from common.utils import expand_path
     ws_root = expand_path(conf().get("agent_workspace", "~/cow"))
@@ -437,6 +446,18 @@ class WebChannel(ChatChannel):
                     "timestamp": time.time(),
                 })
 
+            elif event_type == "agent_cancelled":
+                # Push an explicit cancelled SSE event so the frontend
+                # marks the bubble as stopped. A trailing "done" still
+                # arrives with the partial answer.
+                final_response = data.get("final_response", "")
+                q.put({
+                    "type": "cancelled",
+                    "content": final_response,
+                    "request_id": request_id,
+                    "timestamp": time.time(),
+                })
+
             elif event_type == "agent_end":
                 # Safety net: if the agent finishes with an empty final_response,
                 # chat_channel skips _send_reply (because reply.content is empty),
@@ -756,6 +777,25 @@ class WebChannel(ChatChannel):
             # desire_rtype concept used by other channels).
             is_voice_input = bool(json_data.get('is_voice', False))
 
+            # Fast path for /cancel: bypass the session queue and SSE setup.
+            # Web frontend (stream=true) only listens to SSE, so we return an
+            # inline_reply payload to be rendered synchronously.
+            stripped_prompt = (prompt or "").strip().lower()
+            if stripped_prompt == "/cancel":
+                from agent.protocol import get_cancel_registry
+                cancelled = get_cancel_registry().cancel_session(session_id)
+                lang = (json_data.get('lang') or 'zh').lower()
+                msg_text = _cancel_reply_text(cancelled, lang)
+                logger.info(
+                    f"[WebChannel] /cancel fast-path: session={session_id}, cancelled={cancelled}, lang={lang}"
+                )
+                return json.dumps({
+                    "status": "success",
+                    "request_id": "",
+                    "stream": False,
+                    "inline_reply": msg_text,
+                })
+
             # Append file references to the prompt (same format as QQ channel)
             if attachments:
                 file_refs = []
@@ -862,6 +902,11 @@ class WebChannel(ChatChannel):
                 if itype == "done":
                     post_done = True
                     post_deadline = time.time() + POST_DONE_TAIL_SECONDS
+                elif itype == "cancelled":
+                    # Close SSE tail quickly after cancel; don't wait for the
+                    # full TTS tail since the user already pressed Stop.
+                    post_done = True
+                    post_deadline = time.time() + 3
                 elif itype == "voice_attach":
                     # WSGI buffers the previous chunk until the next yield;
                     # shrink the tail so the generator wakes up quickly to
@@ -872,6 +917,59 @@ class WebChannel(ChatChannel):
         finally:
             self.sse_queues.pop(request_id, None)
 
+    def cancel_request(self):
+        """
+        Cancel an in-flight agent run.
+
+        Body: {"request_id": "...", "session_id": "..."}
+        Either field is sufficient; request_id is preferred when known.
+        Always returns success even when nothing was running, so the
+        client's UX is idempotent.
+        """
+        try:
+            from agent.protocol import get_cancel_registry
+
+            data = web.data()
+            try:
+                json_data = json.loads(data) if data else {}
+            except Exception:
+                json_data = {}
+
+            request_id = (json_data.get("request_id") or "").strip()
+            session_id = (json_data.get("session_id") or "").strip()
+            lang = (json_data.get("lang") or "zh").lower()
+
+            registry = get_cancel_registry()
+            cancelled = 0
+
+            if request_id:
+                if registry.cancel_request(request_id):
+                    cancelled = 1
+
+            if cancelled == 0 and session_id:
+                cancelled = registry.cancel_session(session_id)
+
+            if request_id and request_id in self.sse_queues:
+                self.sse_queues[request_id].put({
+                    "type": "cancelled",
+                    "content": "Cancelled" if lang.startswith("en") else "已中止",
+                    "request_id": request_id,
+                    "timestamp": time.time(),
+                })
+
+            logger.info(
+                f"[WebChannel] cancel request: request_id={request_id!r}, "
+                f"session_id={session_id!r}, cancelled={cancelled}"
+            )
+            return json.dumps({
+                "status": "success",
+                "cancelled": cancelled,
+            })
+
+        except Exception as e:
+            logger.error(f"[WebChannel] cancel_request error: {e}")
+            return json.dumps({"status": "error", "message": str(e)})
+
     def poll_response(self):
         """
         Poll for responses using the session_id.
@@ -967,6 +1065,7 @@ class WebChannel(ChatChannel):
             '/api/voice/tts', 'VoiceTtsHandler',
             '/poll', 'PollHandler',
             '/stream', 'StreamHandler',
+            '/cancel', 'CancelHandler',
             '/chat', 'ChatHandler',
             '/config', 'ConfigHandler',
             '/api/models', 'ModelsHandler',
@@ -1240,6 +1339,12 @@ class PollHandler:
         return WebChannel().poll_response()
 
 
+class CancelHandler:
+    def POST(self):
+        _require_auth()
+        return WebChannel().cancel_request()
+
+
 class StreamHandler:
     def GET(self):
         _require_auth()
diff --git a/channel/wecom_bot/wecom_bot_channel.py b/channel/wecom_bot/wecom_bot_channel.py
index 0fe4500b..ebc1104b 100644
--- a/channel/wecom_bot/wecom_bot_channel.py
+++ b/channel/wecom_bot/wecom_bot_channel.py
@@ -440,6 +440,17 @@ class WecomBotChannel(ChatChannel):
                     state["current"] = ""
                 _push_stream(state, force=True)
 
+            elif event_type == "agent_cancelled":
+                # Flush partial output and strip trailing "---" separator
+                # left over from previous turn, to avoid a dangling divider.
+                if state["current"]:
+                    state["committed"] += state["current"]
+                    state["current"] = ""
+                state["committed"] = state["committed"].rstrip()
+                if state["committed"].endswith("---"):
+                    state["committed"] = state["committed"][:-3].rstrip()
+                _push_stream(state, force=True)
+
         return on_event
 
     # ------------------------------------------------------------------
diff --git a/docs/cli/general.mdx b/docs/cli/general.mdx
index cb3f933d..36af1783 100644
--- a/docs/cli/general.mdx
+++ b/docs/cli/general.mdx
@@ -39,6 +39,14 @@ Mode:    agent
 Session: 12 messages | 8 skills loaded
 ```
 
+## cancel
+
+中止当前会话正在运行的 Agent 任务。在 Agent 执行长时间任务（例如多轮工具调用、长流式输出）时，可随时发送 `/cancel`，Agent 会在下一次工具执行前停止。Web 端、微信、企业微信、飞书等各通道均可使用。
+
+```text
+/cancel
+```
+
 ## config
 
 查看或修改运行时配置。修改后立即生效，无需重启服务。
diff --git a/docs/cli/index.mdx b/docs/cli/index.mdx
index ce67be27..f6462ecb 100644
--- a/docs/cli/index.mdx
+++ b/docs/cli/index.mdx
@@ -57,6 +57,7 @@ Others:
 | --- | --- |
 | `/help` | 显示命令帮助 |
 | `/status` | 查看服务状态和配置 |
+| `/cancel` | 中止当前正在运行的 Agent 任务 |
 | `/config` | 查看或修改运行时配置 |
 | `/skill` | 管理技能（安装、卸载、启用、禁用等） |
 | `/memory dream [N]` | 手动触发记忆蒸馏（默认 3 天，最大 30） |
@@ -82,6 +83,7 @@ Others:
 | version | ✓ | ✓ |
 | status | ✓ | ✓ |
 | logs | ✓ | ✓ |
+| cancel | ✗ | ✓ |
 | config | ✗ | ✓ |
 | context | — | ✓ |
 | memory (子命令) | ✗ | ✓ |
diff --git a/docs/en/cli/general.mdx b/docs/en/cli/general.mdx
index ae78f3e4..8107fcb5 100644
--- a/docs/en/cli/general.mdx
+++ b/docs/en/cli/general.mdx
@@ -25,6 +25,14 @@ View current session and service status, including process info, model configura
 /status
 ```
 
+## cancel
+
+Abort the agent task currently running in this session. When the agent is busy with a long task (e.g. multi-turn tool calls or a long streaming response), send `/cancel` and the agent will stop before the next tool execution. Available across all channels — Web, WeChat, WeCom, Feishu, etc.
+
+```text
+/cancel
+```
+
 ## config
 
 View or modify runtime configuration. Changes take effect immediately without restarting.
diff --git a/docs/en/cli/index.mdx b/docs/en/cli/index.mdx
index 36147261..e13b45a3 100644
--- a/docs/en/cli/index.mdx
+++ b/docs/en/cli/index.mdx
@@ -57,6 +57,7 @@ In the Web console or any connected channel, type `/` to see command suggestions
 | --- | --- |
 | `/help` | Show command help |
 | `/status` | View service status and configuration |
+| `/cancel` | Abort the currently running agent task |
 | `/config` | View or modify runtime configuration |
 | `/skill` | Manage skills (install, uninstall, enable, disable, etc.) |
 | `/memory dream [N]` | Manually trigger memory distillation (default 3 days, max 30) |
@@ -80,6 +81,7 @@ In the Web console or any connected channel, type `/` to see command suggestions
 | version | ✓ | ✓ |
 | status | ✓ | ✓ |
 | logs | ✓ | ✓ |
+| cancel | ✗ | ✓ |
 | config | ✗ | ✓ |
 | context | — | ✓ |
 | memory (subcommands) | ✗ | ✓ |
diff --git a/docs/ja/cli/general.mdx b/docs/ja/cli/general.mdx
index ab24a14e..e31d7cf3 100644
--- a/docs/ja/cli/general.mdx
+++ b/docs/ja/cli/general.mdx
@@ -25,6 +25,14 @@ description: ステータスの確認、設定管理、コンテキスト制御
 /status
 ```
 
+## cancel
+
+現在のセッションで実行中の Agent タスクを中止します。Agent が長時間のタスク（マルチターンのツール呼び出しや長いストリーミング応答など）を実行している間、`/cancel` を送信すると、次のツール実行の前に停止します。Web、WeChat、企業微信、Feishu など、すべてのチャネルで利用可能です。
+
+```text
+/cancel
+```
+
 ## config
 
 実行時設定の表示または変更を行います。変更は即座に反映され、再起動は不要です。
diff --git a/docs/ja/cli/index.mdx b/docs/ja/cli/index.mdx
index f8e28eff..4d00a654 100644
--- a/docs/ja/cli/index.mdx
+++ b/docs/ja/cli/index.mdx
@@ -57,6 +57,7 @@ Web コンソールや接続されたチャネルの会話で `/` を入力す
 | --- | --- |
 | `/help` | コマンドヘルプを表示 |
 | `/status` | サービスの状態と設定を表示 |
+| `/cancel` | 実行中の Agent タスクを中止 |
 | `/config` | 実行時設定の表示・変更 |
 | `/skill` | スキル管理（インストール、アンインストール、有効化、無効化など） |
 | `/memory dream [N]` | 記憶蒸留を手動トリガー（デフォルト 3 日、最大 30） |
@@ -80,6 +81,7 @@ Web コンソールや接続されたチャネルの会話で `/` を入力す
 | version | ✓ | ✓ |
 | status | ✓ | ✓ |
 | logs | ✓ | ✓ |
+| cancel | ✗ | ✓ |
 | config | ✗ | ✓ |
 | context | — | ✓ |
 | memory（サブコマンド） | ✗ | ✓ |
diff --git a/plugins/cow_cli/cow_cli.py b/plugins/cow_cli/cow_cli.py
index fc721f51..b7dc6371 100644
--- a/plugins/cow_cli/cow_cli.py
+++ b/plugins/cow_cli/cow_cli.py
@@ -26,16 +26,19 @@ from common.log import logger
 from cli import __version__
 
 
-# Known top-level subcommands that cow supports
+# Known top-level subcommands that cow supports.
+# "start" / "stop" / "restart" refer to daemon lifecycle on the host shell;
+# in chat, "/cancel" aborts the in-flight agent run instead.
 KNOWN_COMMANDS = {
     "help", "version", "status", "logs",
     "start", "stop", "restart",
+    "cancel",
     "skill", "context", "config",
     "knowledge", "memory",
     "install-browser",
 }
 
-# Commands that can only run from the CLI (terminal), not in chat
+# Commands that can only run from the CLI (terminal), not in chat.
 CLI_ONLY_COMMANDS = {"start", "stop", "restart"}
 
 # Commands that can only run from chat (need access to in-process memory)
@@ -225,6 +228,7 @@ class CowCliPlugin(Plugin):
             "  /help          显示此帮助",
             "  /version       查看版本",
             "  /status        查看运行状态",
+            "  /cancel        中止当前正在运行的 Agent 任务",
             "  /logs [N]      查看最近N条日志 (默认20)",
             "  /context       查看当前对话上下文信息",
             "  /context clear 清除当前对话上下文",
@@ -250,6 +254,41 @@ class CowCliPlugin(Plugin):
     def _cmd_version(self, args: str, e_context, **_) -> str:
         return f"CowAgent v{__version__}"
 
+    # ------------------------------------------------------------------
+    # cancel — abort the in-flight agent run for the current session.
+    # Fallback handler; in practice chat_channel/web_channel intercept
+    # /cancel earlier so it bypasses the per-session serial queue.
+    # ------------------------------------------------------------------
+
+    def _cmd_cancel(self, args: str, e_context: EventContext, session_id: str = "", **_) -> str:
+        """Signal the running agent to halt at its next checkpoint."""
+        from agent.protocol import get_cancel_registry
+
+        target_session = self._get_session_id(e_context, fallback=session_id)
+        registry = get_cancel_registry()
+
+        # Prefer per-turn request_id (matches the key agent_bridge registered)
+        cancelled = 0
+        request_id = ""
+        if e_context is not None:
+            try:
+                ctx = e_context["context"]
+                request_id = ctx.kwargs.get("request_id") or ctx.get("request_id", "")
+            except Exception:
+                request_id = ""
+
+        if request_id and registry.cancel_request(request_id):
+            cancelled = 1
+
+        # Fall back to session-wide cancel
+        if cancelled == 0 and target_session:
+            cancelled = registry.cancel_session(target_session)
+
+        if cancelled <= 0:
+            return "当前没有可中止的任务。"
+
+        return "🛑 已中止"
+
     # ------------------------------------------------------------------
     # status
     # ------------------------------------------------------------------