Shared persistent memory for coding agents, backed by Markdown and hybrid search
Uses Markdown files as the source of truth and Milvus as a rebuildable shadow index. It offers dense vector + BM25 + RRF fusion, deduplication, live sync, and three-stage progressive search (first a small index, then expanded context, and finally transcript details). Keeping memory human-readable, editable, and versionable with Git reduces the risk of cloud/vector lock-in.
Multi-agent teams that want portable, file-based shared project memory.
Setting up Milvus/embeddings for semantic search is heavier than AgentMemory's keyless path; adoption is still low.
/plugin marketplace add zilliztech/memsearch /plugin install memsearch uv tool install memsearch