@ideadesignmedia/memory-mcp

SQLite-backed memory for MCP agents. Ships a CLI and programmatic API.

Highlights (v2)

• Single global memory space; no owner segregation nor types.
• Minimal schema: subject, content, date_created, date_updated, expires_at (embedding is internal-only).
• Optional FTS5 indexing for better search; falls back to LIKE when unavailable.
• Auto-generates embeddings internally via OpenAI when a key is provided; otherwise relies on text search. Embeddings are never accepted from nor returned to MCP clients.

Install / Run

Quick run (no install):

npx -y @ideadesignmedia/memory-mcp --db=/abs/path/memory.db --topk=6

Install locally (dev dependency) and run:

npm i -D @ideadesignmedia/memory-mcp
npx memory-mcp --db=/abs/path/memory.db --topk=6

Other ecosystem equivalents:

• pnpm: pnpm dlx @ideadesignmedia/memory-mcp --db=... --topk=6
• yarn (classic): yarn dlx @ideadesignmedia/memory-mcp --db=... --topk=6

CLI usage

You can invoke it directly (if globally installed) or via npx as shown above.

Optional flags:

• --embed-key=sk-... supply the embedding API key (same as MEMORY_EMBEDDING_KEY).
• --embed-model=text-embedding-3-small override the embedding model (same as MEMORY_EMBED_MODEL).

Codex config example

Using npx so no global install is required. Add to ~/.codex/config.toml:

[mcp_servers.memory]
command = "npx"
args = ["-y", "@ideadesignmedia/memory-mcp", "--db=/abs/path/memory.db", "--topk=6"]

Programmatic API

import { MemoryStore, runStdioServer } from "@ideadesignmedia/memory-mcp";

const store = new MemoryStore("./memory.db");
// All store methods are async
const id = await store.insert({
  ownerId: "user-123",
  type: "preference",
  subject: "favorite color",
  content: "blue",
});

// Run as an MCP server over stdio
await runStdioServer({
  dbPath: "./memory.db",
  defaultTopK: 6,
  embeddingApiKey: process.env.MEMORY_EMBEDDING_KEY, // optional
});

Tools (v2)

• memory-create

  • Create a memory with subject, content. Optionally include ttlDays (to compute expires_at).
  • Response: { ok: true } on success, or { ok: false, error: string } on failure.

• memory-search

  • Search globally by text with optional k (semantic ranking used internally when available).
  • Response items: { id, subject, content }.

• memory-update

  • Update fields of a memory by id: subject, content, ttlDays (recomputes expires_at), or expiresAt.
  • Response: { ok: true } on success, or { ok: false, error: string } on failure.

• memory-delete

  • Delete a memory by id.
  • Response: { ok: true } on success, or { ok: false, error: string } on failure.

• memory-get

  • Fetch a single memory by id.
  • Response item: { id, subject, content, dateCreated, dateUpdated }.

Embeddings

Embeddings are optional—without a key the server relies on text search. Embeddings are internal only; they are never accepted from or returned to clients.

Set MEMORY_EMBEDDING_KEY (or pass --embed-key=... to the CLI) to automatically create embeddings when remembering/importing memories and to embed recall queries. The default model is text-embedding-3-small; override it with MEMORY_EMBED_MODEL or --embed-model. To disable the built-in generator when using the programmatic API, pass embeddingProvider: null to createMemoryMcpServer. To specify a key programmatically, pass embeddingApiKey: "sk-...".

Limits and validation (v2)

• memory-create: subject max 160 chars; content max 2000; optional ttlDays.
• memory-search: optional query max 1000; k up to 20.
• memory-update: accepts partial fields; ttlDays recomputes expires_at.