vecmem

[](LICENSE) [](package.json) [](https://nodejs.org) [](https://www.typescriptlang.org) [](tests/) [](src/mcp/)

> Your notes become AI's memory

You have markdown notes scattered across folders. You know the answer is *somewhere* in there, but where?

vecmem indexes your .md files and lets you (or your AI assistant) search them instantly — not just by keywords, but by meaning.

You: "How did we set up authentication?"
AI:  "I don't have access to your notes."

You: *manually searches 50 files*
You: *finds it 10 minutes later in docs/auth/oauth.md*

vecmem index ./docs                        # One-time: index your notes
vecmem query "how did we set up auth?"     # Instant: finds it in 100ms

Found 5 results (87ms)

┌─ docs/auth/oauth.md ─── score: 0.94 ────────────┐
│ ## OAuth 2.0 Flow                                 │
│ Authentication uses OAuth 2.0 with PKCE.          │
│ The flow starts with a redirect to /auth/login... │
└───────────────────────────────────────────────────┘

It finds oauth.md even though you searched "auth" — because it understands meaning, not just words.

npm install -g vecmem

vecmem init                              # Find all .md files in your project
vecmem index                             # Index them (takes a few seconds)
vecmem query "database schema"           # Search by meaning
vecmem status                            # See what's indexed

vecmem is an [MCP server](http