Unverified50% confidenceBenchmarkExact time
As a control, when using the adapter of another skill, the Executing Writer recovers only 1.3-1.5% of key instructions.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/9/2026
First Seen
Valid until: 2/7/2027
Sources
Related Entities
Related Claims
UnverifiedThe Executing Writer is a small writer model built on top of the adapted model, and on held-out skills it recovers 39-44% of key instructions.69% similarUnverifiedSkills通过按需加载机制，仅在触发时才将skill.md内容载入上下文，平时几乎不占用Token配额67% similarUnverifiedWhen modifying a skill, you usually don't need to invoke the skill creator again; just use a normal prompt to have the model modify the corresponding Markdown file.66% similarUnverifiedAfter rewriting tool definitions following the tool description principle (one tool does one thing, tool names are semantically meaningful), the model's tool invocation accuracy improved from less than half to over 80%.66% similarUnverifiedSkill++'s core method is to rewrite only the name and description of each newly added skill, with no need to train the router or use labeled data, and without modifying the skill body.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1149720API
curl https://kongchang.com/api/v1/knowledge/claims/1149720MCP
get_claim(id=1149720)