Palimpse for Claude
Claude remembers the conversation. Your project needs more.
Claude already gives you Projects with knowledge, a personal memory and a long context inside one conversation. What it does not give you is a project state that survives across conversations, models and teammates. Palimpse adds that layer, connected over MCP.
What Claude keeps. What ends with the chat.
Claude’s built-in memory is real and useful, including memory inside Projects. It is just scoped to your account and to Claude: it holds facts and summaries, not the structured state a project needs to be picked up by someone or something else.
Claude today
- Projects, with knowledge you upload
- Project-scoped memory inside claude.ai Projects
- A personal memory of facts about you
- A long context within a single conversation
What still disappears
- The decisions made in a conversation, and why they were made
- The tasks that were still open when the chat ended
- How the documents changed, and who changed them
- The state of the work when you switch from claude.ai to Claude Code, or hand it to a teammate
AI memory helps an agent remember facts. Project continuity helps a project preserve its state, history and rationale so that someone or something else can pick it back up later.
Start the next conversation briefed.
Open a new conversation, in claude.ai or Claude Code, and ask Claude to pick up the project. Palimpse answers with a briefing: what changed since the last visit, with open tasks on top.
“Pick up the project.”
Open tasks
Finish the onboarding copy. Review the empty states.
Since your last visit
Two sessions were logged and the positioning document was rewritten.
Latest decision
The demo no longer simulates a chat. The conversation was not the product.
Project memory
The product voice is direct and concise. No superlatives.
No pasted summary, no re-explaining. Claude starts from the current state of the project.
Ruled out stays ruled out.
At the end of a session, Claude logs what happened: a summary, the decisions with the alternatives that were rejected, the documents it touched, the tasks it opened or closed. Decisions carry their why and can supersede earlier ones. So when a new conversation drifts toward an idea you already ruled out, the decision is there, with the reason.
Open tasks persist until they are actually closed, and surface at the top of the next briefing. Nothing depends on you remembering to re-mention them.
Decisions preserve the why. Open work preserves the present. Documents preserve the knowledge. Sessions preserve the movement.
Your project does not belong to Claude.
The same project is readable from ChatGPT, Cursor and any compatible MCP client. Project continuity is the ability to resume a project with its current state, history and rationale intact, regardless of which human, AI, tool or session picks it up next. Claude is one door into it, not the container.
Palimpse itself is built this way: Claude and other tools work on the same project, against the same decisions, tasks and documents.
What is project continuity? How Palimpse is built with Palimpse
Connect Claude in two minutes.
Palimpse is a standard MCP server with OAuth. No key to copy around for everyday use.
Copy the URL
Copy your workspace’s MCP URL. It ends in
/api/mcp.Add it to Claude
Paste it into claude.ai or Claude Code as a connector.
Authorize in the browser
Approve access once. Every project in your workspace is now within reach.
Give Claude a project that outlasts the chat.
Connect once over MCP. Every new conversation starts from the current state: decisions, open tasks, documents, history. Free for solo, free for teams during V1.