What is Claude memory?
Claude memory is a feature of the Claude app (web, desktop and mobile) that saves useful details from your conversations and brings them into later chats, so Claude already knows your role, your projects and how you like to work. Alongside it, chat search lets Claude look up your earlier conversations when you refer to them. Both are described in Anthropic's help article Use Claude's chat search and memory to build on previous context.
Claude saves memory as individual topics while you chat, rather than writing a summary after a conversation ends. You can also tell it directly to "remember this."
This is not the same thing as memory in Claude Code. Claude Code reads CLAUDE.md files you write and keeps its own auto memory per project, as covered in the Claude Code memory guide. The settings on this page don't control those files.
Which Claude plans have memory and chat search?
As of October 2026, per Anthropic's help center:
| Feature | Free | Pro, Max | Team, Enterprise |
|---|---|---|---|
| Memory | On by default | On by default | Owner enables it, then each member opts in |
| Search past chats | No | Yes | Yes |
| Incognito chats | Yes | Yes | Yes |
| Import memory from other AI tools | Yes | Yes | Yes |
The help center says "Memory is on by default for Free, Pro, and Max plans on the web, Claude Desktop, and Claude Mobile" (on the latest app version). On Team and Enterprise, an owner turns memory on in Organization settings > Capabilities, and it stays off for each member until they enable it. Owners can't view or edit members' memories, and turning memory off for the organization permanently deletes everyone's memory data.
For plan names and prices, see claude.com/pricing.
How does chat search work?
Chat search lets Claude find and use earlier conversations. Anthropic's help center says "Searching past chats is available to users on paid plans (Pro, Max, Team, and Enterprise plans)" on the web, desktop and mobile apps. It is on by default once it reaches your account.
You use it by asking, for example "What did we decide about the pricing page last week?" The search shows up as a tool call in the conversation, so you can see when Claude looked something up.
Search respects project boundaries. Outside projects, Claude searches your chats that aren't in a project. Inside a project, it searches only that project's conversations. Incognito chats are never searched.
To turn it off, open Settings > Your Claude (or Settings > Memory in the older layout) and switch off Search and reference chats.
How do you view, edit and pause Claude's memory?
Open Settings > Your Claude and select Manage next to Your memory. Everything Claude remembers is listed as topics. Select a topic to read it, use the edit icon to change it, or delete it. You can also manage memory from a chat: ask Claude to remember something, correct a detail, or forget it.
There are two account-wide switches, and they behave very differently:
- Pause memory keeps your existing memory but stops Claude from using it or creating new memories. Unpause and your existing memory is still there, though chats from the paused period aren't added.
- Reset memory "Permanently deletes all memories including project memories." It can't be undone.
To skip memory for a single chat while keeping it in your history, open a new chat, use the + menu and switch Memory off before sending your first message. Claude won't use or add to memory in that chat, and the chat stays searchable from your other conversations.
Two details are easy to miss. First, deleting a conversation doesn't remove memory entries generated from it, so delete the topic too. Second, memory is included when you export your data.
What does Claude not remember?
By default, Claude does not store topics related to personal or sensitive subjects such as health, race, ethnicity, religious beliefs, politics and gender identity. You can opt in with Include sensitive topics in memory in Settings > Your Claude, which applies only from then on. On Team and Enterprise, enabling memory for the organization doesn't enable sensitive topics.
Some details are excluded no matter what. Anthropic's help center says "Some information is never saved to memory, even if you ask," and lists examples such as government ID numbers, financial account numbers and immigration status.
How does memory work in projects?
"Each project has its own memory," per the help center. Inside a project, Claude uses that project's memory alongside your memory from chats outside projects, but never memory from other projects, which keeps client work and side projects apart. To keep a project fully separate from the rest of your account, turn off Use account memory in the project's memory settings. Resetting memory deletes project memories too.
Incognito isn't available inside projects. To keep one project chat out of memory, use the + menu and turn off Memory before your first message. The Claude Projects guide covers project instructions and knowledge, which are better places for anything that must always apply.
Memory is also shared with Claude Cowork when Cowork runs in the cloud. Local Cowork sessions don't use it.
What are incognito chats?
Incognito chats are temporary conversations that aren't saved to your chat history or to memory. Per Anthropic's incognito article, they are available on Free, Pro, Max, Team and Enterprise.
- Starting one: outside a project, click the ghost icon in the upper right of a new chat. The chat gets a black border and an "Incognito chat" label.
- Memory and search: an incognito chat doesn't use existing memory, adds nothing to memory, and is never found by chat search.
- Retention: incognito chats are kept for 30 days by default for safety, or longer under an organization's custom retention setting. They aren't used for model training.
- Limits: you can't convert an incognito chat into a regular one or reopen it after closing. Claude can still use your profile preferences and styles in it. In the new Claude experience, incognito chats open in the previous chat experience, so Claude can't create files or run code in them.
Can you move memory between Claude and other AI tools?
Yes, experimentally. Anthropic's import and export article says memory imports are available on Free, Pro, Max, Team and Enterprise on the web and Claude Desktop. You paste text exported from another assistant and click Add to memory, and Claude stores it as individual entries. Claude may not keep imported personal details unrelated to work, and may not always use what you import.
To export, view your memory under Manage, or ask Claude to write out your memories, then save the text in a file. That file is also a useful starting point for writing standing instructions.
Claude memory vs Claude Code memory vs instructions
| Claude app memory | Claude Code memory | Instructions | |
|---|---|---|---|
| Where it lives | Your Claude account | CLAUDE.md files and auto memory on your machine |
Settings, project instructions or a skill |
| Who writes it | Claude, from your chats | You write CLAUDE.md; Claude Code writes auto memory |
You |
| Scope | Your account, with separate memory per project | A repository, your user profile or your organization | Every chat, one project, or matching tasks |
| Best for | Background that saves repeating yourself | Build commands, conventions, repo rules | Rules that must always apply |
Anthropic's personalization article covers the instruction options in the app, and the Claude Code docs cover CLAUDE.md. For repository rules, start with the CLAUDE.md guide.
Keep what must not drift in text you control
Memory is good at background and bad at guarantees: Claude chooses what to save, and you can lose it with one reset. Kipt keeps prompts and Agent Skills as Markdown in a private GitHub repo you own, with Cmd+K search and blanks such as {{client:Acme}} that you fill in when you copy. For procedures you repeat, the free SKILL.md generator turns a prompt into a skill in your browser, with no sign-up.