Core concepts

Memory

A memory is the atomic unit — one structured, versioned piece of context. It might be a decision from a conversation, a snippet of code, a link with a note, or a fact about a project. Every memory has:

• Source attribution (which client or capture method it came from).
• Tags and an optional collection assignment.
• A timestamped history of edits.

See Memory for the full reference.

Collection

A collection is a bucket of related memories. Think of it as a folder or a tag — memories can belong to multiple collections, and you can share or export a collection as a unit. Typical collections: "Project Atlas", "Client conversations", "Decisions log".

Connector

A connector is anything that reads from or writes to your vault. There are two kinds:

• MCP connectors — AI clients (Claude, ChatGPT, Cursor, VS Code) that talk to OmegaVault over the Model Context Protocol.
• Knowledge connectors — destinations that OmegaVault pushes structured memories to (Notion, Obsidian).

See Integrations for the full list.

MCP — Model Context Protocol

MCP is the open standard that OmegaVault speaks natively. Instead of building a separate integration for every AI tool, OmegaVault exposes a single MCP server — any compatible client connects to it. The protocol defines how a client discovers resources, calls tools, and reads structured context.

Practically: connect once, recall from anywhere. When MCP becomes the standard transport, your vault stops being tied to any one vendor.

Putting it together

You capture memories from conversations. You organize them into collections. Connectors read those memories over MCP and write new ones back. The whole system is vendor-agnostic — your data is yours, and you can export or delete it at any time.

Next steps

• How OmegaVault works — the capture → structure → sync lifecycle.
• Getting started — capture your first memory now.
• MCP feature page — the protocol, in depth.