Unabyss

Unabyssとは？

Unabyss is a context-management platform for people using several AI assistants and work apps. It connects email, documents, calendars and repositories, extracts their information into structured context and updates it as the sources change. Connected MCP clients can look up that context in natural language, run deeper synthesis and save selected durable memory back to the same layer.

Create an account and add Unabyss to a supported MCP client. In the guided setup, connect at least two sources, choose retrieval permissions and set instructions for what the client should save back. The workspace also supports Markdown exports for context you want to reuse outside a live connection.

Unabyssでできること

Connected, structured context

Pull information from work sources and organize it by topic, source and sensitivity. Ongoing synchronization brings new source information into the context layer.

MCP retrieval and memory updates

Supported clients can query stored context, request multi-step synthesis and persist durable text. These operations let different connected agents use the same maintained information.

Per-connection retrieval controls

Set private-information, company-confidential and source-app exclusions for each credential. Two client connections can receive different allowed views of the same account's context.

Markdown context exports

Generate, read or refresh Markdown exports of selected context. Export generation can run in the background; clients read it when the result is ready.

利用シーン

“Prepare a weekly project update”

Retrieve current notes, calendar items and project information so an assistant can draft an update from connected work context.

“Carry project decisions into another agent”

Use stored conventions, decisions and repository context when switching between supported assistants or coding tools.

“Give tools different context views”

Allow one connection to use company context while applying more restrictive private, confidential or app-level filters to another.

よくある質問

Which sources and assistants can I use?

Connect sources such as Gmail, Notion, Google Calendar, GitHub and Linear, and retrieve their context through supported MCP clients such as Claude, Cursor and ChatGPT. Availability depends on the specific integration.

How do the plans differ?

Plans start with a seven-day trial without a credit card. Pro allows up to three connected MCP agents and 20 connected accounts. Max has unlimited agent/account connections. Team adds a shared workspace and central seat management with a minimum of three seats. MCP rate and usage limits apply separately.

Can I decide what gets saved or retrieved?

Guided setup offers save-everything, save-important or save-nothing instructions for what a client should write back. These are client memory preferences. Each credential can separately exclude private information, company-confidential material or selected source apps from retrieval. MCP access requires cloud-storage mode.

What can I export?

Create Markdown exports of context by topic or preset, save Markdown supplied by a client, or refresh an existing export. Generated exports may take background processing; a ready export can then be read and reused.

Unabyssの代替品

LUCI

luci.memories.ai

LUCI is a local desktop memory tool for Mac and Windows that captures computer activity and meetings, then makes that context available to search, chat, and connected AI coding agents. It is designed for people and teams that need persistent context across calls, screens, notes, and everyday workflows.

Pieces

pieces.app

An on-device work memory layer that helps individuals and teams recover context across their tools.