Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified50% confidenceFactExact time

Skill分为能力型Skill和偏好型Skill，能力型随模型变强会被淘汰，偏好型更持久需要用评估保护防止性能倒退

1

Sources

50%

Confidence

Long-term

Relevance

7/19/2026

First Seen

Sources

别再盲发Skills！DeepMind工程师的Agent评估实战指南

Related Claims

UnverifiedSkill的正确使用方式是渐进式优化：使用→发现问题→加规则→再测试，而非追求一次完美73% similar Unverified通用性与专业性之间存在张力，专注单一任务的工具通常能达到更高精度和可控性，而通用系统面临每个环节够用但不够好的风险73% similar Unverified能力提升型技能存在时效性问题，随着模型版本升级（如未来的Claude Opus 5），某些能力提升技能可能不再需要72% similar Unverified当 Skill 数量膨胀到一定程度，Skill 路由从简单的选择问题演变为检索问题72% similar Unverified能力越强的Agentic模型，行为对齐的挑战也越大70% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/563160

API

curl https://kongchang.com/api/v1/knowledge/claims/563160

MCP

get_claim(id=563160)