How it works

A small program runs quietly on your computer and looks after a folder of memories. When your assistant needs to remember or recall something, it asks that program. The assistant never gets your whole vault: only the few notes that answer the question it's working on, with the date and source attached.

If you point it at a folder of your own documents, it asks for your confirmation in person, on that computer, before it reads anything. A change proposed from anywhere else doesn't take effect on its own.

Where things live

Your assistant

Claude, Cursor, or anything else that speaks MCP. Asks questions; gets a handful of notes back.

↓ ↑

ModelBrain, on your computer

Does the searching locally. Only it is allowed to write to the vault.

↓ ↑

Your vault

Files on your disk, in a location you chose. Back it up like anything else you care about.

Changing your mind, and deleting things

A correction

"Actually, we moved the launch to March." The new note replaces the old one for future answers, and the old one is kept as history.

A revert

"Go back to what I said before the correction." Forgetting the corrected note does this by default: the current version is removed and the earlier one becomes current again. You can instead remove just that version or the whole correction history. Nothing is invented in between.

A deletion

"Forget that document." All of it goes, including every piece of it the system had broken out, not just the paragraph you mentioned.

From here down it gets technical Nothing above depends on reading this.

Local vault → MCP server → any compatible host

Hosts

Any MCP client over stdio. The thin client forwards fourteen tool names and no others.

remember · recall · forget · expand
list_memories · diagnostics · usage · ping
partition_create · partition_list · partition_use
partition_rename · partition_archive · partition_restore

Transport

One local socket (a Unix socket, or a named pipe on Windows), checked against the OS peer identity. Nothing administrative is reachable from a model.

adding a folder happens locally
never over MCP, no override

Daemon + vault

Single writer over append-only segments. Embeddings via llama.cpp, in-process. Full-text and vector indexes over the same objects.

index: ULID → (segment, offset)

One memory, one object

Objects, tags and edges, searched by full text and vector similarity together rather than vector-only. visibility is a real field in the record header that the model can never set; connector provenance is written at ingestion and is not model-writable either.

id           ULID, 16 raw bytes
body         type, text, block_type,
             heading_path, table_data,
             source_uri, sync_connector_id
visibility   header enum (never model-set)
parent_id    chunk → document grouping
tags         inline list per object
edges        inline adjacency (supersedes)

Layered storage, and which layers exist

Layer

Role

Status

Memory

Active context, hot indexes.

in v1

Regular disk

Hot segments, uncompressed and read with ordinary file reads. Not encrypted by ModelBrain; your operating system's disk encryption covers them.

in v1

Compressed disk

Memories that go cold move here once it is enabled (experimental, off by default), zstd-compressed against a dictionary trained on your own notes rather than a generic one, so old context stops piling up disk space without ever leaving your machine.

experimental

Private and local-only blocks are removed from the server-eligible set before placement runs. Privacy is a hard filter, never a cost term. And moving a block between layers never changes what a memory means: eviction is not forgetting.