How OmegaVault works

The three stages

OmegaVault is not a chat log. Each memory enters as raw context, is structured into a reusable unit, then synced back out to every client you've connected. The cycle repeats every time you have a conversation.

1. Capture

Capture is how conversations, notes, and links become memories. Three ways:

• Automatic — connected clients (Claude, ChatGPT) emit conversation events; OmegaVault ingests them in the background.
• Manual — type or paste a note into the Capture screen.
• Programmatic — POST to the API or write over MCP from any client.

See Capture for the details.

2. Structure

Raw context is useless without structure. OmegaVault parses each new memory, extracts entities, and slots it into a memory record with tags, source attribution, and timestamps. Memories live in collections — project or topic buckets you control.

3. Sync

Once structured, your memory is available to every client you've authorized. Sync runs over MCP — when a connected client asks a question, OmegaVault semantically retrieves the relevant memories and returns them in context. Two-way sync means changes made in one place propagate everywhere.

Why MCP matters

The Model Context Protocol is OmegaVault's native transport. Because we speak MCP, we don't ship a separate adapter for each AI client — any MCP-compatible tool can read and write the same vault. Connect once, recall everywhere.

Memory lifecycle

conversation       OmegaVault          every connected
  captured   ───▶   structures  ───▶   client sees it
   & stored          & indexes          on next recall

    Capture  ──────────────── Structure ──────────────── Sync

Next steps

• Core concepts — memories, collections, connectors, MCP.
• Getting started — capture your first memory.
• Sync — how memories flow between clients.