Guide
How Claude's memory works, and how to control it
Claude saves memory as individual topics while you chat, and you can read, edit or delete each one under Settings › Your Claude. Memory is on by default on the Free, Pro, Max and Ultra plans. Deleting a chat does not remove what Claude remembered from it.
What does Claude remember?
Claude adds topics during a conversation, and you can also ask it to remember something directly. Topics cover your role, projects, the people you work with, your communication and technical preferences, and ongoing work.
Sensitive subjects such as health, politics and religion are not saved unless you turn on "Include sensitive topics in memory". Some information is never saved, even with that setting on, including government ID numbers, criminal history and immigration status.
Memory is different from chat search, which looks through past conversations when they are relevant. Search is available on paid plans, and it has its own switch, "Search and reference chats".
How do you view or change it?
Go to Settings › Your Claude, select Manage next to Your memory, and open Topics. Each topic can be read, edited or deleted there. You can also tell Claude in a chat to remember, change or forget something, and the change applies from your next conversation. On older accounts the same controls may appear under Settings › Memory.
Does deleting a chat delete the memory?
No. Memory generated from a conversation stays after the conversation is deleted, so it has to be removed separately, from Topics or by asking Claude to forget it. This matters most when a chat contained something you did not want kept.
What is the difference between pausing, resetting and incognito?
| Option | What it does | What remains |
|---|---|---|
| Pause | Claude stops using memory and adding to it | Existing memory is kept; chats during the pause are not added later |
| Reset | Deletes all memories, including project memories | Nothing, and it cannot be undone |
| Memory off for one chat | That chat neither uses memory nor adds to it | The chat stays in history and can still be found by search |
| Incognito chat | Not saved to history or memory, and not searched later | On Team and Enterprise, it follows retention policies and is kept at least 30 days |
To turn memory off for one chat, click + in a new chat and switch off Memory before the first message; the choice locks once the chat starts. Incognito starts from the ghost icon on a new chat and is not available inside a project.
How does memory work in projects?
Each project has its own memory. Inside a project Claude uses that project's memory and your account memory from chats outside projects, but not the memory of other projects. To keep a project fully separate, turn off "Use account memory" on its page. Deleting a project deletes its memory.
Does it work the same on Team and Enterprise?
Not quite. Owners turn memory on or off for the whole organization, and each member starts with it off. Owners cannot view or edit an individual's memories. Turning memory off for the organization permanently deletes everyone's memory.
Can you move memory in or out?
Yes, as text. To bring memory in, go to Settings › Your Claude, select Start import, paste what your other assistant exported, and select Add to memory. Conversation history cannot be imported. Anthropic describes imports as experimental, and personal details unrelated to work may not be kept. To take memory out, open Manage next to Your memory, or ask Claude to write out your memories exactly as stored. The steps for moving between assistants are in the switching guide.
Does Claude Code share this memory?
Anthropic describes this memory as shared between chat and Claude Cowork; it does not list Claude Code. Claude Code has its own mechanisms, CLAUDE.md files and auto memory, kept as files on your machine. The Claude Code memory guide explains them.
Sources, checked on
- Claude Help: Chat search and memory
- Claude Help: Import and export your memory
- Claude Help: Use incognito chats
- Claude blog: Claude's memory works everywhere
Where Coredence fits
Claude's memory is built from what you say in Claude. Coredence is built from your Claude Code sessions, commits and notes, and each memory stays tied to the traces it grew from. It keeps these in a file on your machine rather than in an account.
The Claude app can receive them through its own import: Coredence copies your strongest memories, opens Claude, and you paste. Later handoffs can carry only the memories formed since the last one. Claude Code receives the same memories as a skill, loaded in every session.
Coredence runs on macOS 14 or later (Apple silicon) and Windows 10 or later, with no Coredence account. Consolidation needs a model, and you choose it: your Claude Code subscription, your own API key, or a local model through Ollama. On those paths Coredence itself is free to use. It is not open source, and its own paid plans are not on sale yet.
Try it in your browser