GitHub · Triage and routing · Coding and context
zilliztech/memsearch
Agents that already keep memory and want a cheap filter over what they recall.
What it does
Gives coding agents a cross-platform semantic memory that survives sessions, shipped as plugins for several harnesses. Memory results can be reranked by Jev through the hosted API, with no local model to download.
The repository's own description: “A persistent, unified memory layer for all your AI agents (e.g. Claude Code, Codex, DSH), backed by Markdown and Milvus.”
How it works
Memories are embedded and stored, and a search returns candidates; when reranking is enabled, those candidates go to Jev as one request that scores their relevance, and the order is rewritten before the agent sees it.
The repository, by the numbers
Read from the GitHub API on 2026-09-24. Stars and the last push move daily — quote them with the date, the way we do.
What we checked
- the README and the configuration documentation it links, read 2026-09-25
- the licence file
- the star count and last push, read from the GitHub search API the same day
What we did not check. Whether the reranker improves retrieval on any given corpus. It is an optional switch in a config file, and the repository does not publish an evaluation of it.