Agent Memory Vector Search FAQs

Question 1

When should I use vector search instead of standard keyword search?

Accepted Answer

Use vector search for conceptual queries, synonyms, or when you don't know the exact terms used in a previous discussion. It is ideal for 'find related topics' requests.

Question 2

What is hybrid search in this skill?

Accepted Answer

Hybrid search combines keyword-based BM25 matching with vector similarity (semantic) using Reciprocal Rank Fusion to provide the most relevant results from both technical and conceptual perspectives.

Question 3

Does this skill require a running background process?

Accepted Answer

Yes, the memory-daemon must be running and the vector index must be available for searches to function correctly. You can check status with the memory-daemon status command.

Question 4

What does a similarity score of 0.8 mean?

Accepted Answer

Scores are normalized between 0.0 and 1.0. A score of 0.7 or higher typically indicates a strong semantic match, while 0.5-0.7 represents a moderate conceptual relationship.

Question 5

Can I adjust the balance between keyword and semantic results?

Accepted Answer

Yes, the hybrid search command allows you to tune the --bm25-weight and --vector-weight to favor either exact technical terms or conceptual similarity.

Agent Memory Vector Search

主要功能

使用场景

Agent Memory Vector Search

主要功能

使用场景