Limits

What Onadit does not do today, stated plainly.

Platform#

Mac only, with Apple silicon and macOS 14 or later. There is no Intel, Windows or Linux build.

Capture reads Claude Code and Codex sessions. See Capture for the sources that are opt-in.
Capture starts on the day you turn it on. Sessions that existed before are not imported, unless you resume them.
Capture needs Search by meaning to be on. Without an embedding key, the capture loop does not start.
An imported conversation is dated by its import, not by when it took place.
Chat apps come in through an export file that you request from the provider. Onadit cannot read the ChatGPT or Claude desktop apps directly.

Only MCP clients can search Onadit. Importing a ChatGPT export puts those chats in Onadit. It does not let the ChatGPT app search Onadit.
An agent decides when to call Onadit. Without a line in your rule file, it may not.

A passage an agent recalls becomes part of its conversation, so it reaches that agent's provider.
The archive is not walled off by project. cwd narrows documents and puts the current project's conversations first in a context pack, but conversations from other projects can still come back, and rift_search does not filter conversations by cwd.
Capture summarises sessions through your own Claude Code or Codex, so session text goes to that provider and uses some of your plan. We have not published how much.
Search by meaning, when on, sends passages and questions to Voyage. Keyword search sends nothing.

We have no measured figure for tokens saved or for answer quality.
The speed figures come from 16 fixed tasks on 1 Mac. See Performance for the conditions.

Last updated 12 October 2026. This page is also available as Markdown, and every page is listed in llms.txt.