feat(task): 添加任务修订功能和超时处理机制

添加了 `revise_task` 路由动作类型，允许用户修改、纠正或重新执行最新活动任务结果。
实现了工具失败指导原则，防止相同类别工具重复失败。
为任务规划器添加了超时处理机制，避免长时间等待。

BREAKING CHANGE: 任务路由逻辑已更新，新增 `revise_task` 动作类型。

fix(api): 修复任务详情API返回完整流程投影

修复了任务详情API端点，现在会包含过滤后的流程运行、事件和工件信息，
并确保时间戳字段正确序列化。

refactor(engine): 优化任务技能解析器摘要节点处理

改进了任务技能解析器对摘要节点的处理逻辑，对于仅依赖文本生成功能的摘要节
点不再分配具体技能，直接使用依赖项输出进行汇总。

test: 增加任务修订和超时处理测试用例

添加了测试用例验证任务修订输入记录反馈、超时回退到单模式以及
摘要节点技能解析等新功能。

app-instance/backend/beaver/engine/loop.py

@@ -20,6 +20,13 @@ from beaver.tools import ToolContext
 from .loader import EngineLoader, EngineLoadResult
 
 
+TOOL_FAILURE_GUIDANCE_PROMPT = (
+    "# Tool Failure Guidance\n\n"
+    "If the same class of tools fails repeatedly in a run, stop retrying with query variants. "
+    "Use available materials, state uncertainty clearly, and provide partial confirmed results."
+)
+
+
 @dataclass(slots=True)
 class AgentProfile:
     """Runtime profile for a Beaver agent instance."""
@@ -548,6 +555,7 @@ class AgentLoop:
                     parent_session_id=parent_session_id,
                 ),
                 execution_context=execution_context,
+                extra_sections=[TOOL_FAILURE_GUIDANCE_PROMPT],
             )
             context_result = context_builder.build_messages(build_input)
             if skill_selection_context:

app-instance/backend/beaver/engine/session/models.py

@@ -75,6 +75,8 @@ class MessageRecord:
             "role": self.role,
             "content": self.content,
         }
+        if self.timestamp is not None:
+            payload["timestamp"] = self.timestamp
         if self.run_id:
             payload["run_id"] = self.run_id
         if self.event_payload:

app-instance/backend/beaver/interfaces/web/app.py

@@ -1635,6 +1635,8 @@ def create_app(
 
     @app.get("/api/tasks/{task_id}")
     async def get_task(task_id: str, request: Request) -> dict[str, Any]:
+        from beaver.services.process_service import SessionProcessProjector
+
         loaded = get_agent_service(request).create_loop().boot()
         task_service = loaded.task_service
         if task_service is None:
@@ -1642,10 +1644,18 @@ def create_app(
         task = task_service.get_task(task_id)
         if task is None:
             raise HTTPException(status_code=404, detail="Task not found")
+        process_projection = SessionProcessProjector(
+            loaded.session_manager,
+            loaded.run_memory_store,
+        ).project(task.session_id)
+        filtered_process = _filter_task_process_projection(process_projection, task_id)
         return {
             **task_service.to_api_dict(task),
             "events": [event.to_dict() for event in task_service.list_events(task_id)],
             "runs": _task_run_views(task, task_service.list_events(task_id), loaded.session_manager, loaded.run_memory_store),  # type: ignore[arg-type]
+            "process_runs": filtered_process["runs"],
+            "process_events": filtered_process["events"],
+            "process_artifacts": filtered_process["artifacts"],
         }
 
     @app.delete("/api/tasks/{task_id}")
@@ -2153,6 +2163,33 @@ def _task_run_views(task: Any, events: list[Any], session_manager: Any, run_memo
     return views
 
 
+def _filter_task_process_projection(projection: dict[str, Any], task_id: str) -> dict[str, list[dict[str, Any]]]:
+    def belongs_to_task(item: dict[str, Any]) -> bool:
+        metadata = item.get("metadata")
+        return isinstance(metadata, dict) and metadata.get("task_id") == task_id
+
+    def with_task_metadata(item: dict[str, Any]) -> dict[str, Any]:
+        copied = dict(item)
+        metadata = dict(copied.get("metadata") or {})
+        metadata.setdefault("task_id", task_id)
+        copied["metadata"] = metadata
+        return copied
+
+    runs = [with_task_metadata(item) for item in projection.get("runs", []) if isinstance(item, dict) and belongs_to_task(item)]
+    run_ids = {str(item.get("run_id")) for item in runs if item.get("run_id")}
+    events = [
+        with_task_metadata(item)
+        for item in projection.get("events", [])
+        if isinstance(item, dict) and (belongs_to_task(item) or str(item.get("run_id")) in run_ids)
+    ]
+    artifacts = [
+        with_task_metadata(item)
+        for item in projection.get("artifacts", [])
+        if isinstance(item, dict) and (belongs_to_task(item) or str(item.get("run_id")) in run_ids)
+    ]
+    return {"runs": runs, "events": events, "artifacts": artifacts}
+
+
 def _agent_labels_for_task_events(events: list[Any]) -> dict[str, str]:
     labels: dict[str, str] = {}
     for event in events:

app-instance/backend/beaver/services/agent_service.py

@@ -581,8 +581,96 @@ class AgentService:
             if active_task is None or decision.starts_new_task
             else active_task
         )
+        if active_task is not None and decision.action == "revise_task" and task.task_id == active_task.task_id:
+            task = self._record_revision_feedback_for_task(
+                loaded,
+                task=task,
+                session_id=session_id,
+                comment=message,
+            )
         return await self._run_task_mode(message, runner=runner, kwargs=kwargs, task=task)
 
+    def _record_revision_feedback_for_task(
+        self,
+        loaded: Any,
+        *,
+        task: TaskRecord,
+        session_id: str,
+        comment: str,
+    ) -> TaskRecord:
+        """Mark the latest feedback-eligible run as revised before continuing a task."""
+
+        if task.status not in {"awaiting_feedback", "needs_revision"}:
+            return task
+        run_id = next((item for item in reversed(task.run_ids) if item), None)
+        if not run_id:
+            return task
+
+        existing = next((item for item in task.feedback if item.get("run_id") == run_id), None)
+        if existing is not None:
+            if existing.get("feedback_type") != "revise":
+                return task
+            updated = task
+            already_recorded = True
+        else:
+            task_service = self._require_loaded(loaded, "task_service")
+            updated = task_service.add_feedback(
+                task.task_id,
+                feedback_type="revise",
+                comment=comment,
+                run_id=run_id,
+            )
+            already_recorded = False
+
+        session_manager = self._require_loaded(loaded, "session_manager")
+        session_manager.update_latest_assistant_event_payload(
+            session_id,
+            run_id,
+            {
+                "task_id": updated.task_id,
+                "task_status": updated.status,
+                "feedback_state": "revise",
+            },
+        )
+        if already_recorded:
+            return updated
+
+        session_manager.append_message(
+            session_id,
+            run_id=run_id,
+            role="system",
+            event_type="task_feedback_recorded",
+            event_payload={
+                "task_id": updated.task_id,
+                "feedback_type": "revise",
+                "comment": comment,
+                "task_status": updated.status,
+                "auto_recorded": True,
+            },
+            content=comment,
+            context_visible=False,
+        )
+        validation = ValidationResult.from_dict(updated.validation_result)
+        run_memory_store = self._require_loaded(loaded, "run_memory_store")
+        run_memory_store.update_run_record(
+            run_id,
+            success=False,
+            feedback={
+                "feedback_type": "revise",
+                "comment": comment,
+                "task_status": updated.status,
+            },
+        )
+        run_memory_store.update_skill_effects_for_run(
+            run_id,
+            success=False,
+            feedback_score=self._feedback_score_for_learning("revise", validation),
+            notes=comment.strip() or "revise",
+        )
+        skill_learning_service = self._require_loaded(loaded, "skill_learning_service")
+        skill_learning_service.rescore_skill_versions()
+        return updated
+
     async def _run_task_mode(
         self,
         message: str,
@@ -1018,7 +1106,11 @@ class AgentService:
                     if plan.final_synthesis_instruction
                     else None
                 ),
-                "Use the team outputs as internal evidence. Produce the final user-facing answer yourself.",
+                (
+                    "Use successful team outputs as internal evidence. If one or more nodes failed, "
+                    "do not blindly repeat failed tool calls. Produce a user-visible fallback answer "
+                    "with available evidence and clearly state any missing or uncertain data."
+                ),
             ]
             if item
         )
@@ -1031,7 +1123,11 @@ class AgentService:
                 f"Planner reason: {plan.reason}",
                 f"Strategy: {plan.graph.strategy if plan.graph else ''}",
                 f"Error: {error}",
-                "Proceed as the main agent and produce the best possible final answer.",
+                (
+                    "Proceed as the main agent. Do not blindly repeat failed tool calls; "
+                    "produce a user-visible fallback answer with available evidence and clearly "
+                    "state any missing or uncertain data."
+                ),
             ]
         )
 

app-instance/backend/beaver/skills/builtin/intent-agent-router/SKILL.md

@@ -13,6 +13,7 @@ Your only job is to classify the current user message into one routing decision:
 
 - `simple_chat`
 - `continue_task`
+- `revise_task`
 - `new_task`
 - `close_task`
 - `abandon_task`
@@ -27,6 +28,23 @@ Choose `new_task` when the user asks for anything that needs the main Task agent
 
 The Intent Agent has no tools. If a request needs a tool, do not apologize and do not say you cannot access it. Route it to Task mode so the main agent can use tools.
 
+When there is an active task, do not force every new user message into that task. Use the active task and recent conversation to decide:
+
+- Choose `revise_task` when the user asks to change, correct, refine, expand, reformat, or redo the latest active task result.
+- Choose `continue_task` for neutral follow-up questions or additional next steps that still belong to the active task.
+- Choose `new_task` when the user asks for clearly unrelated work.
+- Choose `close_task` when the user says the task is satisfactory or finished, such as "可以了", "就这样", or "that's good".
+- Choose `abandon_task` when the user says to stop, cancel, or no longer do the active task.
+
+Examples with an active weather task:
+
+- "再详细一点" -> `revise_task`
+- "加上明后天穿衣建议" -> `revise_task`
+- "顺便查一下深圳" -> `continue_task`
+- "帮我写一个采购合同" -> `new_task`
+- "可以了" -> `close_task`
+- "不用了" -> `abandon_task`
+
 ## Must Create Task
 
 Choose `new_task` when there is no active task and the request asks to:

app-instance/backend/beaver/tasks/planner.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import asyncio
 import json
 from dataclasses import dataclass, field
 from typing import Any, Literal
@@ -77,6 +78,7 @@ class TaskExecutionPlanner:
         attempt_index: int,
         latest_validation: ValidationResult | None = None,
         provider_bundle: ProviderBundle | None = None,
+        timeout_seconds: float = 30.0,
     ) -> TaskExecutionPlan:
         provider = None
         model = None
@@ -87,29 +89,32 @@ class TaskExecutionPlanner:
         if provider is None:
             return TaskExecutionPlan.single("planner_provider_unavailable")
         try:
-            response = await provider.chat(
-                messages=[
-                    {
-                        "role": "system",
-                        "content": (
-                            "You choose whether an internal Beaver Task attempt should run as a single "
-                            "main-agent pass or use a small sub-agent team first. Return only compact JSON."
-                        ),
-                    },
-                    {
-                        "role": "user",
-                        "content": self._prompt(
-                            task=task,
-                            user_message=user_message,
-                            attempt_index=attempt_index,
-                            latest_validation=latest_validation,
-                        ),
-                    },
-                ],
-                tools=None,
-                model=model,
-                max_tokens=4096,
-                temperature=0.0,
+            response = await asyncio.wait_for(
+                provider.chat(
+                    messages=[
+                        {
+                            "role": "system",
+                            "content": (
+                                "You choose whether an internal Beaver Task attempt should run as a single "
+                                "main-agent pass or use a small sub-agent team first. Return only compact JSON."
+                            ),
+                        },
+                        {
+                            "role": "user",
+                            "content": self._prompt(
+                                task=task,
+                                user_message=user_message,
+                                attempt_index=attempt_index,
+                                latest_validation=latest_validation,
+                            ),
+                        },
+                    ],
+                    tools=None,
+                    model=model,
+                    max_tokens=4096,
+                    temperature=0.0,
+                ),
+                timeout=timeout_seconds,
             )
             plan = self.from_json(response.content or "")
             return await self._resolve_plan(
@@ -120,7 +125,9 @@ class TaskExecutionPlanner:
                 provider_bundle=provider_bundle,
             )
         except Exception as exc:
-            return TaskExecutionPlan.single("planner_failed", fallback_error=str(exc))
+            detail = str(exc)
+            error = f"{type(exc).__name__}: {detail}" if detail else type(exc).__name__
+            return TaskExecutionPlan.single("planner_failed", fallback_error=error)
 
     async def _resolve_plan(
         self,

app-instance/backend/beaver/tasks/router.py

@@ -69,6 +69,14 @@ class MainAgentRouter:
         reason = str(payload.get("reason") or raw_action or "llm_router")
         short_title = _clean_short_title(payload.get("short_title") or payload.get("title"))
 
+        if raw_action in {"revise_task", "revise", "revision", "needs_revision"}:
+            return MainAgentDecision(
+                mode="task",
+                reason=reason,
+                starts_new_task=active_task is None,
+                short_title=short_title,
+                action="revise_task" if active_task is not None else "create_task",
+            )
         if raw_action in {"continue_task", "continue", "task"}:
             return MainAgentDecision(
                 mode="task",
@@ -146,13 +154,16 @@ class MainAgentRouter:
             "Actions:\n"
             "- simple_chat: no Task should be created or continued.\n"
             "- continue_task: keep the user in the active Task.\n"
+            "- revise_task: user asks to change, correct, refine, expand, reformat, or redo the latest active Task result.\n"
             "- new_task: start a separate new Task.\n"
             "- close_task: user explicitly says the active Task is done/satisfactory/finished.\n"
             "- abandon_task: user explicitly says to stop, cancel, abandon, or no longer do the active Task.\n\n"
             "Critical policy:\n"
-            "- If there is an active Task, choose continue_task unless the user's topic is completely unrelated "
+            "- If there is an active Task, choose continue_task or revise_task unless the user's topic is completely unrelated "
             "to that Task or the user explicitly closes/abandons it.\n"
-            "- Follow-up questions, corrections, partial changes, extra constraints, and result discussion stay in continue_task.\n"
+            "- Choose revise_task when the active Task is awaiting feedback or needs revision and the user asks for changes "
+            "such as '改一下', '加上', '删除', '换成', '再详细点', '格式改成', '不要', or equivalent wording.\n"
+            "- Choose continue_task for neutral follow-up questions or additional next steps that do not imply dissatisfaction with the previous result.\n"
             "- Use new_task only when the user clearly asks to start a different task.\n"
             "- If there is no active Task, choose new_task only for work that requires execution, iteration, tools, files, "
             "implementation, validation, or multi-step completion. Otherwise choose simple_chat.\n"

app-instance/backend/beaver/tasks/skill_resolver.py

@@ -93,6 +93,29 @@ class TaskSkillResolver:
             for item in node.agent.metadata.get("required_capabilities", [])
             if str(item).strip()
         ]
+        if self._is_summary_only_node(node, skill_query=skill_query, required_capabilities=required_capabilities):
+            resolved = self._generic_node(
+                node,
+                pinned_skill_names=[],
+                pinned_skill_contexts=[],
+                metadata={
+                    **node.agent.metadata,
+                    "skill_query": skill_query,
+                    "required_capabilities": required_capabilities,
+                    "selected_skill_names": [],
+                    "ephemeral_skill_names": [],
+                    "summary_uses_dependency_outputs_only": True,
+                },
+            )
+            return resolved, SkillResolutionReport(
+                node_id=node.node_id,
+                skill_query=skill_query,
+                required_capabilities=required_capabilities,
+                selected_skill_names=[],
+                ephemeral_used=False,
+                reason="summary node uses dependency outputs directly",
+            )
+
         selected = await self._select_published_skills(
             query="\n".join(
                 part
@@ -226,6 +249,34 @@ class TaskSkillResolver:
                 selected.append(name)
         return selected
 
+    @staticmethod
+    def _is_summary_only_node(
+        node: ExecutionNode,
+        *,
+        skill_query: str,
+        required_capabilities: list[str],
+    ) -> bool:
+        node_id = node.node_id.strip().lower()
+        query = skill_query.strip().lower()
+        capabilities = {item.strip().lower() for item in required_capabilities}
+        task_text = node.task.strip().lower()
+        summary_identity = node_id in {"summarize", "summary", "synthesis"} or query in {
+            "summarization",
+            "summary",
+            "synthesis",
+            "final synthesis",
+        }
+        text_only_capabilities = not capabilities or capabilities.issubset(
+            {"text generation", "summarization", "summary", "synthesis"}
+        )
+        dependency_summary_task = (
+            "summary" in task_text
+            or "summarize" in task_text
+            or "synthesis" in task_text
+            or "compile" in task_text
+        )
+        return summary_identity and text_only_capabilities and dependency_summary_task
+
     @staticmethod
     def _generic_node(
         node: ExecutionNode,
@@ -246,7 +297,9 @@ class TaskSkillResolver:
                 },
             ),
             inherited_pinned_skills=pinned_skill_names,
-            inherited_pinned_skill_contexts=list(pinned_skill_contexts or node.inherited_pinned_skill_contexts),
+            inherited_pinned_skill_contexts=list(
+                node.inherited_pinned_skill_contexts if pinned_skill_contexts is None else pinned_skill_contexts
+            ),
         )
 
     @staticmethod

app-instance/backend/tests/unit/test_active_task_api.py

@@ -78,3 +78,81 @@ def test_task_delete_api_removes_backend_task(tmp_path: Path) -> None:
     assert deleted.json()["task_id"] == task.task_id
     assert all(item["task_id"] != task.task_id for item in listed.json())
     assert missing.status_code == 404
+
+
+def test_task_detail_api_includes_filtered_process_projection(tmp_path: Path) -> None:
+    service = AgentService(workspace=tmp_path)
+    loaded = service.create_loop().boot()
+    task = loaded.task_service.create_task(  # type: ignore[union-attr]
+        session_id="web:detail",
+        description="补充赛事数据",
+    )
+    other_task = loaded.task_service.create_task(  # type: ignore[union-attr]
+        session_id="web:detail",
+        description="不相关任务",
+    )
+    loaded.session_manager.append_message(
+        "web:detail",
+        role="system",
+        event_type="task_execution_planned",
+        event_payload={
+            "task_id": task.task_id,
+            "attempt_index": 2,
+            "plan_mode": "team",
+            "strategy": "parallel",
+            "node_ids": ["search_match_result", "search_match_stats"],
+            "reason": "needs separate evidence gathering",
+        },
+        context_visible=False,
+    )
+    loaded.session_manager.append_message(
+        "web:detail",
+        role="system",
+        event_type="task_team_run_failed",
+        event_payload={
+            "task_id": task.task_id,
+            "attempt_index": 2,
+            "plan_mode": "team",
+            "strategy": "parallel",
+            "team_success": False,
+            "team_run_ids": ["sub-run"],
+            "node_results": [
+                {
+                    "node_id": "search_match_stats",
+                    "success": False,
+                    "output_text": "",
+                    "run_id": "sub-run",
+                    "finish_reason": "max_tool_iterations",
+                    "error": "max_tool_iterations",
+                }
+            ],
+            "error": "one or more team nodes failed",
+        },
+        context_visible=False,
+    )
+    loaded.session_manager.append_message(
+        "web:detail",
+        role="system",
+        event_type="task_execution_planned",
+        event_payload={
+            "task_id": other_task.task_id,
+            "attempt_index": 1,
+            "plan_mode": "single",
+            "strategy": None,
+            "node_ids": [],
+        },
+        context_visible=False,
+    )
+    app = create_app(service=service, manage_service_lifecycle=False)
+
+    with TestClient(app) as client:
+        response = client.get(f"/api/tasks/{task.task_id}")
+
+    assert response.status_code == 200
+    payload = response.json()
+    assert [run["run_id"] for run in payload["process_runs"]] == [
+        f"task:{task.task_id}:attempt:2",
+        "sub-run",
+    ]
+    assert {event["actor_name"] for event in payload["process_events"]} == {"Task Planner", "Task Team", "search_match_stats"}
+    assert all(event["metadata"]["task_id"] == task.task_id for event in payload["process_events"])

app-instance/backend/tests/unit/test_main_agent_router.py

@@ -103,6 +103,20 @@ def test_router_continues_active_task_from_llm_decision() -> None:
     assert provider.calls[0]["max_tokens"] == 256
 
 
+def test_router_marks_revision_from_llm_decision() -> None:
+    decision = asyncio.run(
+        MainAgentRouter().classify(
+            "再详细一点，并加上表格",
+            active_task=_task(),
+            provider=RouterProvider('{"action":"revise_task","reason":"user requested changes","short_title":"任务连续性"}'),
+        )
+    )
+
+    assert decision.is_task
+    assert decision.starts_new_task is False
+    assert decision.action == "revise_task"
+
+
 def test_router_receives_thinking_mode() -> None:
     provider = RouterProvider('{"action":"simple_chat","reason":"simple"}')
     decision = asyncio.run(

app-instance/backend/tests/unit/test_session_message_model.py

@@ -0,0 +1,12 @@
+from beaver.engine.session.models import MessageRecord
+
+
+def test_conversation_message_preserves_timestamp() -> None:
+    record = MessageRecord(
+        role="user",
+        content="hello",
+        timestamp=1_779_329_600.0,
+        message_id=42,
+    )
+
+    assert record.to_conversation_message()["timestamp"] == 1_779_329_600.0

app-instance/backend/tests/unit/test_task_execution_planner.py

@@ -27,6 +27,22 @@ class PlannerProvider(LLMProvider):
         return "stub-model"
 
 
+class HangingPlannerProvider(LLMProvider):
+    async def chat(
+        self,
+        messages: list[dict],
+        tools: list[dict] | None = None,
+        model: str | None = None,
+        max_tokens: int = 4096,
+        temperature: float = 0.7,
+    ) -> LLMResponse:
+        await asyncio.sleep(10)
+        return LLMResponse(content='{"mode":"team"}', finish_reason="stop", provider_name="stub", model="stub-model")
+
+    def get_default_model(self) -> str:
+        return "stub-model"
+
+
 def _task() -> TaskRecord:
     return TaskRecord(
         task_id="task-1",
@@ -49,6 +65,13 @@ def _bundle(response: str) -> ProviderBundle:
     )
 
 
+def _hanging_bundle() -> ProviderBundle:
+    return ProviderBundle(
+        main_runtime=SimpleNamespace(model="stub-model", provider_name="stub"),
+        main_provider=HangingPlannerProvider(),
+    )
+
+
 def test_planner_selects_single_mode() -> None:
     plan = asyncio.run(
         TaskExecutionPlanner().plan(
@@ -95,6 +118,22 @@ def test_planner_builds_team_graph() -> None:
     assert plan.final_synthesis_instruction == "merge the findings"
 
 
+def test_planner_timeout_falls_back_to_single() -> None:
+    plan = asyncio.run(
+        TaskExecutionPlanner().plan(
+            task=_task(),
+            user_message="implement workflow",
+            attempt_index=1,
+            provider_bundle=_hanging_bundle(),
+            timeout_seconds=0.01,
+        )
+    )
+
+    assert plan.mode == "single"
+    assert plan.reason == "planner_failed"
+    assert "TimeoutError" in (plan.fallback_error or "")
+
+
 def test_planner_team_nodes_can_target_skills_without_agent_roles() -> None:
     plan = TaskExecutionPlanner().from_json(
         """

app-instance/backend/tests/unit/test_task_mode_feedback.py

@@ -290,6 +290,109 @@ def test_active_task_continues_until_llm_closes_it(tmp_path: Path) -> None:
     assert loaded.task_service.active_task_view("web:continue") is None
 
 
+def test_active_task_revision_input_records_feedback_and_reruns(tmp_path: Path) -> None:
+    service = AgentService(
+        loader=EngineLoader(
+            workspace=tmp_path,
+            task_execution_planner=_single_planner(),
+            validation_service=StubValidationService(
+                [
+                    ValidationResult(passed=True, score=0.9, validator="test"),
+                    ValidationResult(passed=True, score=0.95, validator="test"),
+                ]
+            ),
+        )
+    )
+
+    first = asyncio.run(
+        service.process_direct(
+            "查询珠海天气",
+            session_id="web:revise-direct",
+            provider_bundle=_bundle("珠海天气概览", route_action="new_task"),
+        )
+    )
+    second = asyncio.run(
+        service.process_direct(
+            "再详细一点，并加上明后天穿衣建议",
+            session_id="web:revise-direct",
+            provider_bundle=_bundle("更新后的珠海天气和穿衣建议", route_action="revise_task"),
+        )
+    )
+    loaded = service.create_loop().boot()
+    task = loaded.task_service.get_task(first.task_id)
+    messages = loaded.session_manager.get_messages_as_conversation(first.session_id)
+    first_assistant = [
+        message
+        for message in messages
+        if message.get("role") == "assistant" and message.get("run_id") == first.run_id
+    ][-1]
+    user_messages = [message.get("content") for message in messages if message.get("role") == "user"]
+
+    assert second.task_id == first.task_id
+    assert task is not None
+    assert task.status == "awaiting_feedback"
+    assert len(task.run_ids) == 2
+    assert task.feedback == [
+        {
+            "feedback_type": "revise",
+            "comment": "再详细一点，并加上明后天穿衣建议",
+            "run_id": first.run_id,
+            "created_at": task.feedback[0]["created_at"],
+        }
+    ]
+    assert first_assistant["feedback_state"] == "revise"
+    assert "再详细一点，并加上明后天穿衣建议" in user_messages
+
+
+def test_explicit_revision_feedback_then_input_reruns_without_duplicate_feedback(tmp_path: Path) -> None:
+    service = AgentService(
+        loader=EngineLoader(
+            workspace=tmp_path,
+            task_execution_planner=_single_planner(),
+            validation_service=StubValidationService(
+                [
+                    ValidationResult(passed=True, score=0.9, validator="test"),
+                    ValidationResult(passed=True, score=0.95, validator="test"),
+                ]
+            ),
+        )
+    )
+
+    first = asyncio.run(
+        service.process_direct(
+            "查询珠海天气",
+            session_id="web:explicit-revise",
+            provider_bundle=_bundle("珠海天气概览", route_action="new_task"),
+        )
+    )
+    feedback = asyncio.run(
+        service.submit_feedback(
+            session_id=first.session_id,
+            run_id=first.run_id,
+            feedback_type="revise",
+            comment="准备补充穿衣建议",
+        )
+    )
+    second = asyncio.run(
+        service.process_direct(
+            "加上明后天穿衣建议",
+            session_id="web:explicit-revise",
+            provider_bundle=_bundle("更新后的珠海天气和穿衣建议", route_action="revise_task"),
+        )
+    )
+    loaded = service.create_loop().boot()
+    task = loaded.task_service.get_task(first.task_id)
+
+    assert feedback["task_status"] == "needs_revision"
+    assert second.task_id == first.task_id
+    assert task is not None
+    assert task.status == "awaiting_feedback"
+    assert len(task.run_ids) == 2
+    assert len(task.feedback) == 1
+    assert task.feedback[0]["feedback_type"] == "revise"
+    assert task.feedback[0]["comment"] == "准备补充穿衣建议"
+
+
 def test_validation_failure_retries_once(tmp_path: Path) -> None:
     service = AgentService(
         loader=EngineLoader(
@@ -545,6 +648,8 @@ def test_task_mode_team_failure_still_uses_main_synthesis(tmp_path: Path) -> Non
     assert result.output_text == "fallback synthesized answer"
     assert any(event.event_type == "task_team_run_failed" for event in events)
     assert "sub-agent unavailable" in main_provider.calls[0][0]["content"]
+    assert "same class of tools fails repeatedly" in main_provider.calls[0][0]["content"]
+    assert "user-visible fallback answer" in main_provider.calls[0][0]["content"]
 
 
 def test_task_mode_team_retry_hides_first_synthesis_run(tmp_path: Path) -> None:

app-instance/backend/tests/unit/test_task_skill_resolver.py

@@ -65,8 +65,8 @@ def _publish_skill(workspace: Path, *, skill_name: str) -> None:
     store = SkillSpecStore(workspace)
     draft = DraftService(store).create_new_skill_draft(
         skill_name=skill_name,
-        proposed_content="# API Contract Review\n\nCheck schema compatibility and breaking changes.",
-        proposed_frontmatter={"description": "API contract compatibility review", "tools": []},
+        proposed_content=f"# {skill_name}\n\nCheck schema compatibility and breaking changes.",
+        proposed_frontmatter={"description": f"{skill_name} capability", "tools": []},
         created_by="tester",
         reason="test",
     )
@@ -174,3 +174,51 @@ def test_task_skill_resolver_generates_ephemeral_guidance_when_missing(tmp_path:
     assert reports[0].ephemeral_guidance_id is not None
     assert reports[0].ephemeral_guidance_name == "api-compatibility-review"
     assert reports[0].ephemeral_used is True
+
+
+def test_task_skill_resolver_keeps_summary_nodes_skillless(tmp_path: Path) -> None:
+    _publish_skill(tmp_path, skill_name="multi-search-engine")
+    provider = RecordingProvider(['["multi-search-engine"]'])
+    resolver = TaskSkillResolver(
+        skills_loader=SkillsLoader(tmp_path),
+        draft_service=DraftService(SkillSpecStore(tmp_path)),
+    )
+    graph = ExecutionGraph(
+        strategy="dag",
+        nodes=[
+            ExecutionNode(
+                "summarize",
+                "Compile a clear, concise summary from dependency outputs for the user.",
+                AgentDescriptor(
+                    name="summarize",
+                    metadata={
+                        "skill_query": "Summarization",
+                        "required_capabilities": ["text generation"],
+                    },
+                ),
+                depends_on=["verify_result"],
+                inherited_pinned_skills=["multi-search-engine"],
+                inherited_pinned_skill_contexts=[
+                    SkillContext(name="ephemeral:search-guidance", content="Search again.")
+                ],
+            )
+        ],
+    )
+
+    resolved, reports = asyncio.run(
+        resolver.resolve_graph(
+            graph,
+            task=_task(),
+            user_message="summarize result",
+            attempt_index=2,
+            provider_bundle=_bundle(provider),
+        )
+    )
+
+    assert resolved.nodes[0].inherited_pinned_skills == []
+    assert resolved.nodes[0].inherited_pinned_skill_contexts == []
+    assert resolved.nodes[0].agent.metadata["selected_skill_names"] == []
+    assert reports[0].selected_skill_names == []
+    assert reports[0].ephemeral_used is False
+    assert reports[0].reason == "summary node uses dependency outputs directly"
+    assert provider.calls == []

app-instance/frontend/app/(app)/notifications/[scheduledRunId]/page.tsx

@@ -150,6 +150,7 @@ export default function NotificationDetailPage() {
           selectedRunId={null}
           onSelectRun={() => {}}
           onFeedback={() => {}}
+          onRequestRevision={() => {}}
         />
       </div>
 

app-instance/frontend/app/(app)/page.tsx

@@ -18,41 +18,12 @@ import {
   uploadFile,
   wsManager,
 } from '@/lib/api';
+import { mergeServerWithPendingUsers } from '@/lib/chat-messages';
 import { pickAppText } from '@/lib/i18n/core';
 import { useAppI18n } from '@/lib/i18n/provider';
 import { useChatStore } from '@/lib/store';
 import type { ActiveTask, ChatMessage, FileAttachment, SessionUpdatedEvent, WsEvent } from '@/types';
 
-function messageFingerprint(msg: ChatMessage): string {
-  const attachmentKey = (msg.attachments ?? [])
-    .map((a) => `${a.file_id ?? ''}:${a.name}:${a.content_type}:${a.size ?? ''}`)
-    .join('|');
-  return `${msg.role}::${String(msg.content)}::${attachmentKey}`;
-}
-
-function mergeServerWithPendingUsers(serverMessages: ChatMessage[], localMessages: ChatMessage[]): ChatMessage[] {
-  const counts = new Map<string, number>();
-  for (const message of serverMessages) {
-    const key = messageFingerprint(message);
-    counts.set(key, (counts.get(key) ?? 0) + 1);
-  }
-
-  const pendingUsers: ChatMessage[] = [];
-  for (const message of localMessages) {
-    const key = messageFingerprint(message);
-    const count = counts.get(key) ?? 0;
-    if (count > 0) {
-      counts.set(key, count - 1);
-      continue;
-    }
-    if (message.role === 'user') {
-      pendingUsers.push(message);
-    }
-  }
-
-  return [...serverMessages, ...pendingUsers];
-}
-
 function isSessionUpdatedEvent(data: WsEvent | Record<string, unknown>): data is SessionUpdatedEvent {
   return data.type === 'session_updated' && typeof data.session_id === 'string';
 }
@@ -101,11 +72,13 @@ export default function ChatPage() {
   const [thinkingModeEnabled, setThinkingModeEnabled] = useState(loadThinkingModePreference);
   const [pendingFiles, setPendingFiles] = useState<Array<{ file: File; id?: string; progress: number; error?: string }>>([]);
   const [activeTask, setActiveTask] = useState<ActiveTask | null>(null);
+  const [revisionTargetRunId, setRevisionTargetRunId] = useState<string | null>(null);
   const messagesEndRef = useRef<HTMLDivElement>(null);
   const messageViewportRef = useRef<HTMLDivElement>(null);
   const textareaRef = useRef<HTMLTextAreaElement>(null);
   const fileInputRef = useRef<HTMLInputElement>(null);
   const loadSessionReqSeq = useRef(0);
+  const loadedSessionIdRef = useRef<string | null>(null);
   const refreshSessionOnReconnectRef = useRef(false);
   const hasConnectedRef = useRef(false);
   const shouldSnapToLatestRef = useRef(true);
@@ -185,10 +158,15 @@ export default function ChatPage() {
   }, [loadActiveTask, setIsLoading, setIsThinking, setMessages, setSessionProcess]);
 
   useEffect(() => {
-    clearMessages();
-    setIsLoading(false);
-    setIsThinking(false);
+    const didSwitchSession = loadedSessionIdRef.current !== null && loadedSessionIdRef.current !== sessionId;
+    loadedSessionIdRef.current = sessionId;
+    if (didSwitchSession) {
+      clearMessages();
+      setIsLoading(false);
+      setIsThinking(false);
+    }
     setActiveTask(null);
+    setRevisionTargetRunId(null);
     void loadSessionMessages(sessionId);
     void loadActiveTask(sessionId);
   }, [clearMessages, loadActiveTask, loadSessionMessages, sessionId, setIsLoading, setIsThinking]);
@@ -304,10 +282,33 @@ export default function ChatPage() {
       size: item.file.size,
     }));
 
+    const msgContent = text || pickAppText(locale, '（仅附件）', '(Attachments only)');
+
+    if (revisionTargetRunId && text) {
+      setIsLoading(true);
+      setIsThinking(false);
+      updateMessageFeedback(revisionTargetRunId, 'revise');
+      try {
+        await submitChatFeedback({
+          sessionId,
+          runId: revisionTargetRunId,
+          feedbackType: 'revise',
+          comment: msgContent,
+        });
+      } catch (err: any) {
+        setIsThinking(false);
+        setIsLoading(false);
+        updateMessageFeedback(revisionTargetRunId, undefined, err?.message || pickAppText(locale, '反馈提交失败', 'Feedback failed'));
+        return;
+      } finally {
+        setRevisionTargetRunId(null);
+      }
+    } else {
+      setRevisionTargetRunId(null);
+    }
+
     setInput('');
     setPendingFiles([]);
-
-    const msgContent = text || pickAppText(locale, '（仅附件）', '(Attachments only)');
     addMessage({
       role: 'user',
       content: msgContent,
@@ -371,7 +372,7 @@ export default function ChatPage() {
         });
       }
     }
-  }, [addMessage, input, isLoading, loadActiveTask, loadSessionMessages, loadSessions, locale, pendingFiles, sessionId, setIsLoading, setIsThinking, setSessionProcess, thinkingModeEnabled]);
+  }, [addMessage, input, isLoading, loadActiveTask, loadSessionMessages, loadSessions, locale, pendingFiles, revisionTargetRunId, sessionId, setIsLoading, setIsThinking, setSessionProcess, thinkingModeEnabled, updateMessageFeedback]);
 
   const handleFeedback = useCallback(async (runId: string, feedbackType: 'satisfied' | 'revise' | 'abandon', comment?: string) => {
     updateMessageFeedback(runId, feedbackType);
@@ -391,6 +392,11 @@ export default function ChatPage() {
     }
   }, [loadActiveTask, loadSessionMessages, loadSessions, locale, sessionId, setSessionProcess, updateMessageFeedback]);
 
+  const handleRequestRevision = useCallback((runId: string) => {
+    setRevisionTargetRunId(runId);
+    textareaRef.current?.focus();
+  }, []);
+
   const handleKeyDown = (e: React.KeyboardEvent) => {
     if (e.key === 'Enter' && !e.shiftKey && !e.nativeEvent.isComposing) {
       e.preventDefault();
@@ -426,6 +432,7 @@ export default function ChatPage() {
     setSessionId(id);
     setSelectedRunId(null);
     setActiveTask(null);
+    setRevisionTargetRunId(null);
     clearMessages();
     useChatStore.getState().resetProcessState();
     try {
@@ -444,6 +451,7 @@ export default function ChatPage() {
       if (key === sessionId) {
         setSessionId('web:default');
         setActiveTask(null);
+        setRevisionTargetRunId(null);
         clearMessages();
         useChatStore.getState().resetProcessState();
       }
@@ -460,6 +468,7 @@ export default function ChatPage() {
   const handleSelectSession = (key: string) => {
     setSelectedRunId(null);
     setActiveTask(null);
+    setRevisionTargetRunId(null);
     setSessionId(key);
   };
 
@@ -554,24 +563,29 @@ export default function ChatPage() {
             selectedRunId={selectedSessionRunId}
             onSelectRun={(runId) => setSelectedRunId(selectedSessionRunId === runId ? null : runId)}
             onFeedback={handleFeedback}
+            onRequestRevision={handleRequestRevision}
           />
         </div>
 
         <div className="bg-background px-8 pb-8 pt-4">
           <div className="mx-auto max-w-5xl">
-            {activeTask && (
+            {(activeTask || revisionTargetRunId) && (
               <div className="mb-2 flex">
-                <Link
-                  href={`/tasks/${encodeURIComponent(activeTask.task_id)}`}
-                  className="inline-flex max-w-full items-center gap-2 rounded-full border border-[#D8D2CE] bg-[#F7F6F5] px-3 py-1.5 text-xs text-foreground transition-colors hover:bg-[#EFEEED]"
-                  title={activeTask.description}
-                >
-                  <span className="shrink-0 text-muted-foreground">{pickAppText(locale, '当前任务', 'Current task')}:</span>
-                  <span className="truncate font-medium">{activeTask.short_title}</span>
-                  <span className="shrink-0 rounded-full bg-white px-2 py-0.5 text-[11px] text-muted-foreground">
-                    {activeTaskStatusLabel(activeTask.status, locale)}
-                  </span>
-                </Link>
+                {activeTask ? (
+                  <Link
+                    href={`/tasks/${encodeURIComponent(activeTask.task_id)}`}
+                    className="inline-flex max-w-full items-center gap-2 rounded-full border border-[#D8D2CE] bg-[#F7F6F5] px-3 py-1.5 text-xs text-foreground transition-colors hover:bg-[#EFEEED]"
+                    title={activeTask.description}
+                  >
+                    <span className="shrink-0 text-muted-foreground">
+                      {revisionTargetRunId ? pickAppText(locale, '修改任务', 'Revising task') : pickAppText(locale, '当前任务', 'Current task')}:
+                    </span>
+                    <span className="truncate font-medium">{activeTask.short_title}</span>
+                    <span className="shrink-0 rounded-full bg-white px-2 py-0.5 text-[11px] text-muted-foreground">
+                      {revisionTargetRunId ? pickAppText(locale, '待输入修改要求', 'Awaiting revision') : activeTaskStatusLabel(activeTask.status, locale)}
+                    </span>
+                  </Link>
+                ) : null}
               </div>
             )}
             {pendingFiles.length > 0 && (
@@ -607,7 +621,11 @@ export default function ChatPage() {
                 value={input}
                 onChange={(e) => setInput(e.target.value)}
                 onKeyDown={handleKeyDown}
-                placeholder={pickAppText(locale, '今天想聊什么？', 'What would you like to talk about today?')}
+                placeholder={
+                  revisionTargetRunId
+                    ? pickAppText(locale, '请输入修改要求', 'Describe the requested changes')
+                    : pickAppText(locale, '今天想聊什么？', 'What would you like to talk about today?')
+                }
                 rows={1}
                 className="block w-full resize-none border-0 bg-transparent px-2 pb-8 pt-1 text-[17px] leading-7 placeholder:text-muted-foreground focus:outline-none disabled:cursor-not-allowed disabled:opacity-50"
                 style={{ minHeight: '72px', maxHeight: '200px' }}

app-instance/frontend/app/(app)/tasks/[taskId]/page.tsx

@@ -15,7 +15,7 @@ import { pickAppText } from '@/lib/i18n/core';
 import { useAppI18n } from '@/lib/i18n/provider';
 import { buildTaskRuntimeView, type TaskRuntimeNodeView } from '@/lib/task-runtime';
 import { useChatStore } from '@/lib/store';
-import type { BackendTask, BackendTaskRun, ProcessArtifact, ProcessEvent } from '@/types';
+import type { BackendTask, BackendTaskRun, ProcessArtifact, ProcessEvent, ProcessRun } from '@/types';
 
 type TaskFeedbackType = 'satisfied' | 'revise' | 'abandon';
 type TaskFeedbackItem = {
@@ -217,6 +217,8 @@ export default function TaskDetailPage() {
           }
         />
 
+        <BackendExecutionStages task={backendTask} />
+
         <Card>
           <CardHeader>
             <CardTitle className="text-base">{pickAppText(locale, 'Agent 执行过程', 'Agent conversation process')}</CardTitle>
@@ -549,6 +551,80 @@ function Metric({ label, value }: { label: string; value: string }) {
   );
 }
 
+function BackendExecutionStages({ task }: { task: BackendTask }) {
+  const { locale } = useAppI18n();
+  const runs = task.process_runs ?? [];
+  const events = task.process_events ?? [];
+  const eventsByRun = React.useMemo(() => {
+    const map = new Map<string, ProcessEvent[]>();
+    for (const event of events) {
+      map.set(event.run_id, [...(map.get(event.run_id) ?? []), event]);
+    }
+    return map;
+  }, [events]);
+
+  return (
+    <Card>
+      <CardHeader>
+        <CardTitle className="text-base">{pickAppText(locale, '执行阶段', 'Execution stages')}</CardTitle>
+      </CardHeader>
+      <CardContent className="space-y-3">
+        {runs.length === 0 ? (
+          <div className="text-sm text-muted-foreground">{pickAppText(locale, '暂无执行阶段记录', 'No execution stage records yet')}</div>
+        ) : (
+          runs.map((run) => (
+            <BackendProcessRun key={run.run_id} run={run} events={eventsByRun.get(run.run_id) ?? []} />
+          ))
+        )}
+      </CardContent>
+    </Card>
+  );
+}
+
+function BackendProcessRun({ run, events }: { run: ProcessRun; events: ProcessEvent[] }) {
+  const { locale } = useAppI18n();
+  const metadata = run.metadata ?? {};
+  const details = [
+    metadata.attempt_index ? `${pickAppText(locale, '尝试', 'Attempt')} ${String(metadata.attempt_index)}` : null,
+    metadata.plan_mode ? `${pickAppText(locale, '模式', 'Mode')}: ${String(metadata.plan_mode)}` : null,
+    metadata.strategy ? `${pickAppText(locale, '策略', 'Strategy')}: ${String(metadata.strategy)}` : null,
+    metadata.node_id ? `${pickAppText(locale, '节点', 'Node')}: ${String(metadata.node_id)}` : null,
+    metadata.finish_reason ? `${pickAppText(locale, '结束原因', 'Finish')}: ${String(metadata.finish_reason)}` : null,
+  ].filter(Boolean);
+  const error = typeof metadata.error === 'string' && metadata.error ? metadata.error : null;
+
+  return (
+    <div className="rounded-md border border-border bg-background p-3">
+      <div className="flex flex-wrap items-start justify-between gap-3">
+        <div className="min-w-0">
+          <div className="font-medium">{run.title || run.actor_name}</div>
+          <div className="mt-1 text-xs text-muted-foreground">
+            {run.actor_name}
+            {run.started_at ? ` · ${formatTaskRuntimeTime(run.started_at, locale)}` : ''}
+          </div>
+        </div>
+        <TaskRuntimeStatusBadge status={run.status} />
+      </div>
+      {details.length > 0 ? <div className="mt-2 text-xs text-muted-foreground">{details.join(' · ')}</div> : null}
+      {run.summary ? <p className="mt-2 whitespace-pre-wrap text-sm text-muted-foreground">{run.summary}</p> : null}
+      {error ? <p className="mt-2 text-sm text-destructive">{error}</p> : null}
+      {events.length > 0 ? (
+        <div className="mt-3 space-y-2">
+          {events.map((event) => (
+            <div key={event.event_id} className="rounded-md bg-muted/30 px-3 py-2 text-xs">
+              <div className="flex flex-wrap items-center justify-between gap-2">
+                <span className="font-medium">{event.actor_name}</span>
+                <span className="text-muted-foreground">{formatTaskRuntimeTime(event.created_at, locale)}</span>
+              </div>
+              <div className="mt-1 text-muted-foreground">{event.text || event.kind}</div>
+            </div>
+          ))}
+        </div>
+      ) : null}
+    </div>
+  );
+}
+
 function TaskFeedbackPanel({
   sessionId,
   runId,

app-instance/frontend/components/chat-workbench/ChatWorkbench.tsx

@@ -16,6 +16,7 @@ export function ChatWorkbench({
   selectedRunId,
   onSelectRun,
   onFeedback,
+  onRequestRevision,
 }: {
   messages: ChatMessage[];
   isThinking: boolean;
@@ -27,6 +28,7 @@ export function ChatWorkbench({
   selectedRunId: string | null;
   onSelectRun: (runId: string) => void;
   onFeedback: (runId: string, feedbackType: 'satisfied' | 'revise' | 'abandon', comment?: string) => void;
+  onRequestRevision: (runId: string) => void;
 }) {
   return (
     <div className="h-full">
@@ -41,6 +43,7 @@ export function ChatWorkbench({
       selectedRunId={selectedRunId}
       onSelectRun={onSelectRun}
       onFeedback={onFeedback}
+      onRequestRevision={onRequestRevision}
     />
     </div>
   );

app-instance/frontend/components/chat-workbench/MessageList.tsx

@@ -6,6 +6,7 @@ import { Bot, CheckCircle2, ChevronRight, Loader2, Paperclip, RefreshCcw, Thumbs
 
 import type { ChatMessage, ProcessArtifact, ProcessEvent, ProcessRun } from '@/types';
 import { getAccessToken, getFileUrl } from '@/lib/api';
+import { getTaskCardMessageIndexes } from '@/lib/chat-messages';
 import { AgentTeamBlock } from '@/components/chat-workbench/AgentTeamBlock';
 import { MarkdownContent } from '@/components/chat-workbench/MarkdownContent';
 import { ScrollArea } from '@/components/ui/scroll-area';
@@ -40,17 +41,21 @@ function AuthImage({ src, alt, className }: { src: string; alt: string; classNam
 
 function MessageBubble({
   message,
+  showTaskCard,
   canSendFeedback,
   onFeedback,
+  onRequestRevision,
 }: {
   message: ChatMessage;
+  showTaskCard: boolean;
   canSendFeedback: boolean;
   onFeedback: (runId: string, feedbackType: 'satisfied' | 'revise' | 'abandon', comment?: string) => void;
+  onRequestRevision: (runId: string) => void;
 }) {
   const { locale } = useAppI18n();
   const isUser = message.role === 'user';
   const textContent = typeof message.content === 'string' ? message.content : String(message.content || '');
-  const [feedbackMode, setFeedbackMode] = React.useState<'satisfied' | 'revise' | null>(null);
+  const [feedbackMode, setFeedbackMode] = React.useState<'satisfied' | null>(null);
   const [feedbackComment, setFeedbackComment] = React.useState('');
   const validationFailed = message.validation_status === 'failed';
   const validationDetails =
@@ -118,7 +123,7 @@ function MessageBubble({
         ) : (
           <MarkdownContent content={textContent} />
         )}
-        {!isUser && message.task_id && (
+        {!isUser && showTaskCard && message.task_id && (
           <div className="mt-3 rounded-md border border-border bg-muted/35 p-3">
             <div className="flex flex-wrap items-center justify-between gap-3">
               <div className="min-w-0">
@@ -145,7 +150,7 @@ function MessageBubble({
             <p className="mt-2 text-xs leading-5 text-muted-foreground">{validationDetails}</p>
           </details>
         )}
-        {!isUser && canSendFeedback && message.run_id && (
+        {!isUser && (canSendFeedback || message.feedback_state) && message.run_id && (
           <div className="mt-3 space-y-2 border-t border-border/70 pt-3">
             {message.feedback_state ? (
               <div className="flex items-center gap-2 text-xs text-muted-foreground">
@@ -171,7 +176,7 @@ function MessageBubble({
                   </button>
                   <button
                     type="button"
-                    onClick={() => setFeedbackMode('revise')}
+                    onClick={() => onRequestRevision(message.run_id!)}
                     className="inline-flex h-8 items-center gap-1 rounded-md border border-border px-3 text-xs text-muted-foreground hover:bg-accent hover:text-foreground"
                   >
                     <RefreshCcw className="h-3.5 w-3.5" />
@@ -191,11 +196,7 @@ function MessageBubble({
                     <textarea
                       value={feedbackComment}
                       onChange={(event) => setFeedbackComment(event.target.value)}
-                      placeholder={
-                        feedbackMode === 'revise'
-                          ? pickAppText(locale, '写下需要修改的地方...', 'Describe what needs to change...')
-                          : pickAppText(locale, '可选：补充说明...', 'Optional note...')
-                      }
+                      placeholder={pickAppText(locale, '可选：补充说明...', 'Optional note...')}
                       className="min-h-20 w-full resize-none rounded-md border border-input bg-background px-3 py-2 text-sm outline-none focus:ring-2 focus:ring-ring"
                     />
                     <div className="flex justify-end gap-2">
@@ -330,6 +331,7 @@ export function MessageList({
   selectedRunId,
   onSelectRun,
   onFeedback,
+  onRequestRevision,
 }: {
   messages: ChatMessage[];
   isThinking: boolean;
@@ -341,6 +343,7 @@ export function MessageList({
   selectedRunId: string | null;
   onSelectRun: (runId: string) => void;
   onFeedback: (runId: string, feedbackType: 'satisfied' | 'revise' | 'abandon', comment?: string) => void;
+  onRequestRevision: (runId: string) => void;
 }) {
   const { locale } = useAppI18n();
   const visibleMessages = React.useMemo(
@@ -361,6 +364,7 @@ export function MessageList({
       sortTime: parseTimelineTime(message.timestamp) ?? Number.MAX_SAFE_INTEGER / 2 + index,
       order: index,
       message,
+      messageIndex: index,
     }));
     const teamItems = teamGroups.map((group, index) => ({
         kind: 'team' as const,
@@ -377,9 +381,18 @@ export function MessageList({
       return a.order - b.order;
     });
   }, [teamGroups, visibleMessages]);
-  const latestAssistantRunId = [...visibleMessages]
+  const taskCardMessageIndexes = React.useMemo(
+    () => getTaskCardMessageIndexes(visibleMessages),
+    [visibleMessages]
+  );
+  const latestFeedbackRunId = [...visibleMessages]
     .reverse()
-    .find((message) => message.role === 'assistant' && message.run_id && message.task_id)?.run_id;
+    .find((message) =>
+      message.role === 'assistant'
+      && message.run_id
+      && message.task_id
+      && message.task_status === 'awaiting_feedback'
+    )?.run_id;
 
   return (
     <ScrollArea className="h-full px-8" viewportRef={viewportRef}>
@@ -397,8 +410,10 @@ export function MessageList({
             <MessageBubble
               key={item.key}
               message={item.message}
-              canSendFeedback={Boolean(latestAssistantRunId && item.message.run_id === latestAssistantRunId)}
+              showTaskCard={taskCardMessageIndexes.has(item.messageIndex)}
+              canSendFeedback={Boolean(latestFeedbackRunId && item.message.run_id === latestFeedbackRunId)}
               onFeedback={onFeedback}
+              onRequestRevision={onRequestRevision}
             />
           ) : (
             <AgentTeamBlock

app-instance/frontend/lib/chat-messages.test.ts

@@ -0,0 +1,74 @@
+import { describe, expect, it } from 'vitest';
+
+import { getTaskCardMessageIndexes, mergeServerWithPendingUsers } from '@/lib/chat-messages';
+import type { ChatMessage } from '@/types';
+
+describe('chat message helpers', () => {
+  it('keeps pending local user messages after server refreshes', () => {
+    const serverMessages: ChatMessage[] = [
+      {
+        role: 'assistant',
+        content: 'Earlier answer',
+        timestamp: '2026-05-21T08:00:00.000Z',
+      },
+    ];
+    const localMessages: ChatMessage[] = [
+      ...serverMessages,
+      {
+        role: 'user',
+        content: 'Please continue',
+        timestamp: '2026-05-21T08:01:00.000Z',
+      },
+    ];
+
+    expect(mergeServerWithPendingUsers(serverMessages, localMessages)).toEqual([
+      ...serverMessages,
+      localMessages[1],
+    ]);
+  });
+
+  it('does not duplicate local user messages already persisted by the server', () => {
+    const serverMessages: ChatMessage[] = [
+      {
+        role: 'user',
+        content: 'Please continue',
+        timestamp: '2026-05-21T08:01:00.000Z',
+      },
+    ];
+    const localMessages: ChatMessage[] = [
+      {
+        role: 'user',
+        content: 'Please continue',
+        timestamp: '2026-05-21T08:01:01.000Z',
+      },
+    ];
+
+    expect(mergeServerWithPendingUsers(serverMessages, localMessages)).toEqual(serverMessages);
+  });
+
+  it('shows a task card only on the latest assistant message for the same task', () => {
+    const messages: ChatMessage[] = [
+      {
+        role: 'assistant',
+        content: 'I will start.',
+        run_id: 'run-1',
+        task_id: 'task-1',
+      },
+      {
+        role: 'assistant',
+        content: 'I found data.',
+        run_id: 'run-1',
+        task_id: 'task-1',
+      },
+      {
+        role: 'assistant',
+        content: 'Final answer.',
+        run_id: 'run-1',
+        task_id: 'task-1',
+        task_status: 'awaiting_feedback',
+      },
+    ];
+
+    expect(Array.from(getTaskCardMessageIndexes(messages))).toEqual([2]);
+  });
+});

app-instance/frontend/lib/chat-messages.ts

@@ -0,0 +1,44 @@
+import type { ChatMessage } from '@/types';
+
+export function messageFingerprint(msg: ChatMessage): string {
+  const attachmentKey = (msg.attachments ?? [])
+    .map((a) => `${a.file_id ?? ''}:${a.name}:${a.content_type}:${a.size ?? ''}`)
+    .join('|');
+  return `${msg.role}::${String(msg.content)}::${attachmentKey}`;
+}
+
+export function mergeServerWithPendingUsers(serverMessages: ChatMessage[], localMessages: ChatMessage[]): ChatMessage[] {
+  const counts = new Map<string, number>();
+  for (const message of serverMessages) {
+    const key = messageFingerprint(message);
+    counts.set(key, (counts.get(key) ?? 0) + 1);
+  }
+
+  const pendingUsers: ChatMessage[] = [];
+  for (const message of localMessages) {
+    const key = messageFingerprint(message);
+    const count = counts.get(key) ?? 0;
+    if (count > 0) {
+      counts.set(key, count - 1);
+      continue;
+    }
+    if (message.role === 'user') {
+      pendingUsers.push(message);
+    }
+  }
+
+  return [...serverMessages, ...pendingUsers];
+}
+
+export function getTaskCardMessageIndexes(messages: ChatMessage[]): Set<number> {
+  const latestByTask = new Map<string, number>();
+
+  messages.forEach((message, index) => {
+    if (message.role !== 'assistant' || !message.task_id) {
+      return;
+    }
+    latestByTask.set(message.task_id, index);
+  });
+
+  return new Set(latestByTask.values());
+}

app-instance/frontend/types/index.ts

@@ -342,6 +342,9 @@ export interface BackendTask {
   metadata: Record<string, unknown>;
   events?: BackendTaskEvent[];
   runs?: BackendTaskRun[];
+  process_runs?: ProcessRun[];
+  process_events?: ProcessEvent[];
+  process_artifacts?: ProcessArtifact[];
 }
 
 export interface ActiveTask {