How to Reference Another Chat in Claude

To reference another chat in Claude, you have three real options: turn on Claude's memory so it can recall and search past conversations, keep related chats inside a Project that shares context automatically, or manually paste a handoff summary from the old chat into the new one. There is no single "link two chats" button.

Memory, chat search, and Projects behaviour verified 31 July 2026 against the Anthropic Help Center.

That is the honest short answer, and it matters because a lot of guides online imply Claude can magically pull in any old conversation on command. The truth is more nuanced. Whether you can reference another chat in Claude depends on which feature you use, which plan you are on, and whether the two chats share a home. This guide walks through every reliable method, what each one actually does, and when to reach for it. We have used this ourselves, and what follows is shaped by that first-hand experience rather than marketing copy.

Why a new chat doesn't automatically see your other chats

Every conversation is its own sealed context window. A fresh chat starts blank and cannot peek sideways into another thread. That isolation is deliberate: it keeps unrelated topics from bleeding together.

By default, every conversation in Claude is its own sealed container. When you open a fresh chat, Claude starts with a blank slate: it does not carry over what you said in a different conversation, and it cannot "peek" sideways into another thread on its own. Each chat is a separate context window, and once you hit the end of one, the model only knows what is inside it.

This is by design, not a bug. It keeps conversations private from one another and prevents unrelated topics from bleeding together. But it is exactly why people ask how to reference another chat in Claude — they finished a great planning session yesterday and want today's chat to build on it. If you want the deeper mechanics of why a single conversation eventually runs out of room, our piece on the Claude conversation length limit covers how context windows fill up and what happens when they do.

So bridging two chats always means using one of the features below to carry context across that boundary. Let's go through them from most automatic to most manual.

Method 1: Use Claude's memory to reference another chat in Claude

Memory comes in two parts. Memory from chat history carries a synthesized summary of your work into new chats on every plan. Search and reference past chats retrieves specific snippets on request, on paid plans.

Claude's memory feature is the closest thing to automatic cross-chat recall. As of 2026, Anthropic has rolled memory out broadly, and it comes in two related pieces that are easy to confuse:

• Memory from chat history — Claude quietly builds a running, synthesized summary of your preferences, role, and ongoing work across conversations, then carries that forward into new chats without you re-explaining. This is available to all plans, including free.
• Search and reference past chats — an on-demand ability where you explicitly ask Claude to find something you discussed before, and it retrieves relevant snippets from your saved history. This is available on paid plans (Pro, Max, Team, and Enterprise).

Practically, that means you can start a new chat and say, "Find the conversation where we outlined the launch plan and continue from there," and Claude can pull the relevant details in. It uses retrieval behind the scenes, so you will often see it search your history as a visible step.

A few honest caveats. Memory synthesizes rather than storing full transcripts, so fine details from an old chat may not survive perfectly. It is also plan- and rollout-dependent, so exactly what you see can vary. You stay in control: memory and chat search can be toggled in Settings under Capabilities. For a full walkthrough of how this works, what it stores, and how to manage it, see our Claude memory guide. If you are newer to the assistant overall, how to use Claude AI is a good starting point.

Method 2: Keep related chats in a Project (the strongest native answer)

A Project is a workspace holding shared instructions, files, and its own scoped memory. Every chat you start inside it begins with that context loaded, so related conversations share one foundation instead of being bridged after the fact.

If your goal is to keep a body of work coherent — not just reference one old chat once — Projects is the best tool Claude offers. A Project is a dedicated workspace that holds shared instructions, uploaded files, and its own project knowledge. Every chat you start inside that Project begins with that shared context already loaded.

This reframes the whole question. Instead of trying to reference another chat in Claude after the fact, you set things up so all the related chats naturally share the same foundation. Draft five separate conversations about the same client inside one Project, and each of them draws on the same brief, style rules, and reference documents.

Projects also keep memory scoped. Each Project has its own separate memory space and summary, so Claude searches only within that Project and does not mix in chats from elsewhere. That separation is genuinely useful when you juggle multiple clients or products and do not want context cross-contaminating.

It is worth being precise: Projects is not literally "linking" two standalone chats together. It is giving many chats a shared home so they behave as one continuous effort. For most ongoing work, that is what people actually want. Our Claude projects guide covers how to set one up, add knowledge, and get the most out of shared instructions.

Method 3: Manual handoff (reliable, works today, every plan)

Three manual options work everywhere with no rollout uncertainty: ask the old chat for a pasteable handoff summary, copy the few key messages across, or export the whole conversation and upload it as project knowledge.

Automatic features are convenient, but manual methods are the most dependable way to reference another chat in Claude. They work on every plan, right now, with no rollout uncertainty. There are three flavors.

Ask for a handoff summary. In the old chat, type something like: "Write a concise handoff summary of everything important from this conversation that I can paste into a new chat." Claude will produce a tidy recap — decisions made, context, open questions. Copy it, open a fresh conversation, and paste it in as your opening message. This is the single best trick for a clean restart, especially when the original chat has gotten long or cluttered.

Copy the key messages. If you only need a few specific exchanges — for example, the 3 or 4 messages where you settled on an approach — select and copy those, then paste them into the new chat with a line like, "Here's context from an earlier conversation — build on it." Claude reads pasted conversation text just fine and treats it as context.

Export the old chat and feed it in. For a complete record, export the conversation and bring the full text into the new chat or a Project. This is the most thorough option and pairs well with Projects, since you can upload the export as project knowledge. Our guide on how to export Claude chats walks through the steps for saving a conversation.

In our own workflow running this site in Claude Code every day, the handoff summary is the method we lean on most. When a long editorial or debugging session starts to run out of room, we ask Claude to dump the decisions and open threads into a short recap, then paste that straight into a fresh session — it keeps the through-line intact far more reliably than hoping memory synthesized the right details. We learned to be specific in the ask ("include file paths and any TODOs we agreed on"), because a vague request produces a vague summary that loses exactly the concrete bits we needed to reference another chat in Claude for in the first place.

Is there an "@mention another chat" feature?

No. There is no simple mention gesture for injecting an old conversation in the consumer app. What exists instead is memory for recall, chat search for retrieval, and Projects for a shared base.

Short version: no, not as a simple mention in the consumer app the way you might tag a file or a person. People often expect to type an "@" and pick an old conversation to inject — and that specific gesture is not a general feature. What exists instead is the trio above: memory that recalls context, search that retrieves snippets on request, and Projects that share a common base.

It is worth hedging here because Claude ships changes constantly and behavior differs across web, desktop, mobile, and plans. So rather than promising a chat-linking button that may not be in your build, the reliable move is to use memory or Projects for automatic recall and a handoff summary for anything one-off. Anthropic's official help center is the place to confirm exactly what your account supports today, and the Claude site is where new capabilities are announced.

Putting it together: which method should you pick?

Use Projects for ongoing work, a handoff summary for a one-off continuation, and leave memory on for passive continuity. Layering all three is more reliable than depending on any single feature.

Here is the practical rule of thumb. For ongoing work — a project, a client, a book, a long research effort — set up a Project and let shared knowledge do the heavy lifting. You will rarely need to reference another chat in Claude manually because every new chat already starts with the context.

For a one-off — you just want today's chat to build on one earlier conversation — ask the old chat for a handoff summary and paste it into a new one. It takes about 10 seconds and works on any plan.

For passive continuity — you simply want Claude to remember your preferences and current projects without thinking about it — leave memory on and let it accumulate quietly in the background.

You can and should combine them. Many power users run inside a Project, rely on memory for continuity, and still drop a handoff summary at the start of a big new chat to be safe. Layering these gives you the most reliable way to reference another chat in Claude without depending on any single feature.

If you are still deciding whether the assistant fits your workflow at all, what is Claude AI explains the product and its plans in plain terms, which helps you know which memory and search features you actually have access to.

A quick workflow example

End the old chat by asking for a handoff summary, create a Project and paste that summary into its instructions or knowledge, then start every future conversation inside that Project. The thread carries forward automatically.

Say you spent an hour with Claude designing a content calendar. Tomorrow you want to keep going. The clean flow:

1. At the end of today's chat, ask Claude to write a handoff summary.
2. Create a Project called "Content Calendar" and paste that summary into the project instructions or upload it as knowledge.
3. Start every future chat inside that Project.

Now you never have to manually reference another chat in Claude again for this work — the Project carries the thread forward, and memory fills in the softer context around it. That combination is the most durable answer to the whole question.