oneMEM gives AI tools a shared local memory. It turns useful context into compact facts and surfaces the minimum amount of memory sufficient for a query. Every connected agent reads and writes the same SQLite file.

AI agents ───┐
Code editors ┼── MCP ── oneMEM ── ~/.onemem/onemem.db
Local tools ─┘

Install

oneMEM requires Python 3.11 or newer and an API key for any supported language-model provider. Embeddings run locally; there is no embedding service or API key to configure.

uv tool install "onemem[all]"
onemem init

onemem init does the rest:

1. asks which model provider you want to use;
2. recommends a model or lets you enter another model ID;
3. verifies the key and model before saving them;
4. initializes the local database and embedding model;
5. offers background capture; and
6. connects detected AI tools over MCP.

Nothing is uploaded to a oneMEM server. Configuration and credentials stay under ~/.onemem/; the API key is sent only to the provider you select.

Try it

Add something directly:

onemem add "Chose SQLite because it needs zero operations and one-file backups."

Ask for it later:

onemem ask "What storage did I choose, and why?"

Or ask from a connected agent. It receives two MCP tools:

• onemem_recall retrieves relevant memory.
• onemem_log stores something worth remembering.

MCP setup

oneMEM works with any client that can run a local stdio MCP server, including Claude Code, Codex, Cursor, and Windsurf. onemem init automatically configures clients whose command-line tools it detects; other clients only need the onemem-mcp executable path.

claude mcp add --scope user onemem -- "$(command -v onemem-mcp)"
codex mcp add onemem -- "$(command -v onemem-mcp)"

Commands

Where data lives

How retrieval works

session or imported text
        ↓
append-only raw event
        ↓
compact facts + local embeddings + entity anchors
        ↓
deterministic fusion of semantic, keyword, and entity retrieval
        ↓
minimum relevant memory returned to the agent

oneMEM uses an LLM only to interpret and compact language and, optionally, to synthesize the final answer. Storage, indexing, and retrieval are ordinary code.

Benchmarks

Measured on a 100-instance stratified sample of LongMemEval-S:

Development

git clone https://github.com/shashank-tomar0/onemem.git
cd onemem
uv sync --all-extras
uv run pytest -q

License

MIT — Based on Meniscus by magic_bubblez.