About Embedmind
Embedmind is an MCP server in the RAG category: persistent memory for AI agents — one local file, fast, no server. Vector + full-text + graph, via MCP + CLI. Built in Rust.
Install
git clone https://github.com/rodrigoescorsim/embedmindThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (5)
forgetDelete one memory by id (delete-by-query/age is planned).
Parameters not published.
recallHybrid search (vector + full-text + recency, RRF-fused) over everything remembered, best match first with scores. `filters: {key: value}` narrows by metadata, `agent` by writing agent, and `expand_related: true` also pulls each hit's related memories as connected context.
Parameters not published.
relatedNavigate the memory graph: one memory's relations (both directions, with kind, including `supersedes`), or every memory tagged with an entity.
Parameters not published.
rememberStore a memory (text + metadata; embedded and indexed automatically, long text chunked transparently). Optionally tag explicit `entities` and typed `relations` to earlier memories, or `supersedes: [id]` to retire an older memory as a correction. Flags near-duplicates of live, unsuperseded memories at write time.
Parameters not published.
statsFile size, live/forgotten counts, index health, and a per-agent provenance breakdown.
Parameters not published.
Ask AI
Ask AI about Embedmind
Powered by Claude · Grounded in docs
Security checks
6 checks have not been run yet.