Unverified50% confidenceDecision RuleExact time
20–50 input/expected-output pairs should be retained as a test set and run on every prompt change, rather than relying on intuition.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen testing LLM citation behavior, the same prompt should be run at least 20–50 times, ideally expanded to more than 100 times.76% similarUnverifiedIn red-team testing, the scorer was allocated an output budget of only 50 tokens; thinking-type models exhaust all tokens on internal reasoning and ultimately return empty content.72% similarUnverifiedThe number of leads per run is configurable; the demo sets it to 5, but it can be increased to 25, 30, or even 100 per run in practice.71% similarUnverified真实场景中不会因单条反馈更新 ICP，而是在积累 100～200 条反馈后识别出模式再调整70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1004592API
curl https://kongchang.com/api/v1/knowledge/claims/1004592MCP
get_claim(id=1004592)