Unverified50% confidenceFactExact time
Each answer is stored back into FalkorDB as an Episode node with cited sources, the agent does a full-text query over memory before answering, matches are shown as "Remembered from earlier", and memories survive restarts because they live in the graph.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCall records should be kept in their own database, separate from the notes the bot recalls, because if written into the same memory, every recorded answer would become a candidate in the next recall, including the bad answers.67% similarUnverifiedFalkorDB exposes a Cypher interface over the Redis protocol, so HiveMind can retrieve in a single statement the experience, the verification results supporting it, and the historical attempts and failures that produced those results.66% similarUnverifiedRecall should return the original stored records rather than rewritten versions, and return hits together with their sources to enable cross-session continuity.66% similarUnverified语义缓存通过计算新查询与历史查询的向量相似度，当超过阈值时直接返回缓存答案，GPTCache等工具实现了该机制65% similarUnverifiedWhen vector retrieval recalls information, the system converts the current question into a vector, looks up in the database the several historical records with the closest distance, and injects them into the prompt.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1061933API
curl https://kongchang.com/api/v1/knowledge/claims/1061933MCP
get_claim(id=1061933)