Unverified50% confidenceFactExact time
业界通常将Agent的容错行为分为三个等级：静默失败（最差）、抛出异常停止（中等）、识别根因并恢复或上报（最优）
1
Sources
50%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
Related Claims
UnverifiedOptimizing agent evaluation for average success rate, worst-case failure, or expected business damage leads to entirely different engineering decisions.72% similarUnverifiedAgent failure modes can be understood across three architectural layers: planning-layer failure, execution-layer failure, and reflection-layer failure, each with entirely different mitigation strategies.71% similarUnverified在 Skill 或全局配置中提供正例和反例的对比示例（few-shot contrastive examples）能有效纠正 Agent 反复出现的错误68% similarUnverifiedPerplexity categorizes Agent Skill failures into three types: routing issues (a Skill loads when it shouldn't or doesn't load when it should), execution issues (routing is correct but the result is wrong), and hidden conflicts caused by global prompt changes.67% similarUnverifiedFailures in AI agents are often semantic deviations rather than explicit error codes, and this kind of "successful failure" requires the orchestration layer to have semantic-level error recognition capability.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/610717API
curl https://kongchang.com/api/v1/knowledge/claims/610717MCP
get_claim(id=610717)