Claude now remembers you with an editable memory summary and chat search. Basic Memory is a full knowledge base you own. Available to Claude and every other AI tool you use.
Claude’s built-in memory is real memory now: an automatically synthesized summary of your chat history (refreshed daily, viewable and editable in Settings), separate per-project memory, chat search, and team memory on Claude for Work. Available to all Claude users. What it is not: yours, or big. The summary is a digest that lives inside Anthropic’s platform, derives only from your Claude conversations, and works only in Claude. Basic Memory is a knowledge base: thousands of full notes with semantic links, stored as Markdown files you own, readable and writable by Claude over MCP. And by ChatGPT, Cursor, and every other MCP client at the same time. Use both: Claude’s memory for continuity inside Claude, Basic Memory for the knowledge you want to keep, search, and take anywhere.
| Basic Memory | Claude Memory | |
|---|---|---|
| What it is | Knowledge graph in files you own | Synthesized memory summary + project memory + chat search |
| Where knowledge lives | Markdown on your machine or private cloud | Inside Anthropic’s platform |
| You can edit it | Yes. Every note, in any editor | Yes. The summary, in settings or by asking Claude |
| Granularity | Thousands of full notes (50,000/seat on Cloud) | One rolling summary per account + one per project |
| Works across AI vendors | Yes. Claude, ChatGPT, Cursor, any MCP client | No. Claude only |
| Source of knowledge | Conversations, docs, research, imports. Anything | Your Claude conversations |
| Structure | Semantic graph: notes, observations, relations | Rolling summary, refreshed daily |
| Portable / exportable | Yes. Copy the folder | No. Bound to your Claude account |
| Cost | Free local; Cloud from $15/seat/month | Included with Claude |
Claude’s memory has gotten genuinely good. The memory summary. Synthesized from your chats roughly every 24 hours. Is transparent by big-lab standards: you can read it, edit it directly, add instructions, pause it, or reset it. Projects get their own isolated memory, and chat search covers raw history. If you only use Claude and only need continuity, it may be enough.
The structural limit is that it is a digest. One rolling summary (plus one per project) has to compress everything Claude knows about you, so it keeps conclusions and sheds the reasoning, the alternatives you rejected, and the details that did not seem important at the time. Basic Memory keeps the full notes. Decisions, context, research. Linked into a graph that Claude navigates over MCP (it works with Claude Desktop, Claude Code, and claude.ai). The question changes from "what does my summary say?" to "what do my notes say, and what are they connected to?"
Claude’s memory derives from Claude conversations and lives inside Anthropic’s platform. It cannot hold your imported research, your docs, or context from other tools, and it does not come with you if you switch. Basic Memory’s files are yours: use them in ChatGPT tomorrow, grep them in a terminal, edit them in Obsidian, version them in git. If your account is ever locked or you simply change tools, your knowledge comes with you.
For teams, the same distinction applies: Claude’s team memory shares context inside Claude for Work, while Basic Memory Teams gives a team one knowledge base that every member’s tools. And the team’s agents. Can use, with roles and audit logs.
Open source to run locally. Cloud from $15/seat/month with a 7-day trial.
Knowledge tools built to last
© 2026 Basic Memory - Basic Machines