datawhalechina/hello-agents · error · AgentException

Coach Agent执行失败: {str(e)}

Error message

Coach Agent执行失败: {str(e)}

What it means

Outer catch-all in CoachAgent.run: any exception from the four task handlers (_handle_explain_task etc.), from think()/call_tool() inside them, or from _get_user_context is re-wrapped as AgentException with the original message preserved in the suffix. State is set to 'error' before raising. The real cause is always in the appended text; this wrapper adds one layer but no information.

Source

Thrown at Co-creation-projects/Apricity-InnocoreAI/agents/coach.py:66

                result = await self._handle_mimic_task(user_id, content, context)
            elif task_type == "suggest":
                result = await self._handle_suggest_task(user_id, content, context)
            else:
                raise AgentException(f"不支持的任务类型: {task_type}")
            
            self.set_state("completed")
            
            return {
                "status": "success",
                "task_type": task_type,
                "user_id": user_id,
                "result": result,
                "timestamp": datetime.now().isoformat()
            }
            
        except Exception as e:
            self.set_state("error")
            raise AgentException(f"Coach Agent执行失败: {str(e)}")
    
    def get_required_fields(self) -> List[str]:
        """获取必需的输入字段"""
        return ["user_id", "task_type", "content"]
    
    async def _handle_explain_task(self, user_id: str, content: str, context: Dict) -> Dict[str, Any]:
        """处理解释任务"""
        try:
            # 获取用户的历史论文作为上下文
            user_context = await self._get_user_context(user_id)
            
            explain_prompt = f"""
            请用通俗易懂的语言解释以下内容：
            
            需要解释的内容：
            {content}
            
            上下文信息：

View on GitHub (pinned to 606a07d341)

Solutions

  1. Parse the trailing original message — it identifies whether the failure is LLM, DB, or tool.
  2. Re-raise AgentException/TimeoutException unchanged (except (AgentException, TimeoutException): raise) to avoid nested wrappers.
  3. Add raise ... from e to keep the traceback intact.
  4. Fix the root cause in the named handler; this wrapper itself needs no code change once the inner error is resolved.

Example fix

# before
except Exception as e:
    self.set_state("error")
    raise AgentException(f"Coach Agent执行失败: {str(e)}")

# after
except (AgentException, TimeoutException):
    self.set_state("error")
    raise
except Exception as e:
    self.set_state("error")
    raise AgentException(f"Coach Agent执行失败: {e}") from e
Defensive patterns

Strategy: try-catch

Try / catch

try:
    result = await coach.run(input_data)
except AgentException as e:
    root = str(e).replace("Coach Agent执行失败: ", "")
    logger.error("coach failed: %s", root, exc_info=True)
    # classify: LLM/auth errors are ops issues; task-type/field errors are client issues
    raise

Prevention

When it happens

Trigger: think() raising TimeoutException('LLM思考超时') during prompt polish; _get_user_context hitting a database error for an unknown user_id; an inner handler's own except re-raising AgentException, which gets double-wrapped as "Coach Agent执行失败: Coach Agent执行失败: ..." style nesting when handlers follow the same pattern.

Common situations: LLM key/quota problems; DB connection not initialized before coach runs; nested double-wrapping making messages recursive; handlers assuming optional context keys exist.

Related errors

AI-assisted analysis of datawhalechina/hello-agents@606a07d341 (2026-08-14). Data as JSON: /api/errors/d23ab95b560a514b. Report an issue: GitHub.