Unverified50% confidenceSolutionExact time
The prompt requires the model to answer only based on the retrieved documents, cite document IDs, and explicitly return an empty answer when the documents are not helpful; if the output cannot be parsed or the answer is empty, it hands off to a human.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen the retrieval context does not contain the answer, many language models fill the gap with plausible-sounding content, which is exactly the hallucination users notice; fixes include instructing the model to answer only from the provided context, giving it a clear way to opt out, requiring a citation for every claim, setting a retrieval score threshold, and logging rejected questions.74% similarUnverifiedEven when the model is explicitly required to answer only based on retrieved documents, it still fills in gaps using parametric memory and generates seemingly well-formatted citations.72% similarUnverifiedThe fact that the material is on the user's computer does not mean the request has delivered it to the model; information the model never received cannot be conjured up no matter how parameters are adjusted.72% similarUnverifiedThe model was mostly able to ignore irrelevant injections: 10 ordinary questions about push notifications received saved answers for different features, and none of the replies mentioned it.70% similarUnverifiedTool output should return only what the model needs for its next decision, use human-readable values rather than raw nested data, and list-type tools should return a summary and state any truncation.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1066627API
curl https://kongchang.com/api/v1/knowledge/claims/1066627MCP
get_claim(id=1066627)