Unverified50% confidenceSolutionExact time
构建LLM Agent时常见的格式问题包括返回散文而非合法JSON、分镜描述过长、缺失必填字段、凭空捏造不存在的工具参数
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
Unverified在 ReAct 为代表的 Agent 框架中，工具调用的 JSON 格式错误是导致 Agent 循环失败的最常见原因之一75% similarUnverified大语言模型（LLM）启用JSON Mode获取结构化输出时，输出本质上仍是文本，可能出现字段缺失、数据类型错误或模型幻觉导致的非预期字段71% similarUnverifiedAgents that rely on JSON-formatted output may encounter dozens of formatting anomalies across thousands of calls, causing downstream parsing failures.66% similarUnverifiedIt is widely accepted in the industry that JSON is not suitable as a prompt format, because it lacks explicit expression of the meaning of elements within the system.65% similarUnverifiedStructured output constrains format via JSON Schema, which can eliminate randomness at the format level and confine uncertainty to the scope of semantic content.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/571884API
curl https://kongchang.com/api/v1/knowledge/claims/571884MCP
get_claim(id=571884)