What you will be able to do
- Describe what automatic summarisation does, what it requires and what it costs
- Decide when to start a new conversation and when to let a long one continue
- Persist knowledge across chats with projects, memory and past-chat search instead of re-pasting it
1.Letting a long conversation summarise itself
On paid plans, Claude manages long conversations for you. When a chat approaches the context window limit, Claude summarises earlier messages to make room for new content, and the help centre says this lets conversations continue indefinitely in most cases. You might notice Claude "organizing its thoughts" during a long chat. That is automatic context management at work.
The exam tests three facts about it:
1. It needs code execution. Automatic context management only works when code execution is enabled. 2. It isn't free. Longer conversations that trigger it use more of your usage limit. Keeping one chat going indefinitely costs you budget. 3. It doesn't erase the record. Your full chat history is preserved, and Claude can still reference it after summarisation.
It is also not a guarantee. In rare edge cases, such as a very large first message or a system error, you can still hit the context limit.
Checkpoint 1 of 5· Check yourself
Which statement about automatic summarisation in Claude chats is accurate?
Summarisation depends on code execution and costs extra usage. History is preserved and the window size doesn't change.
“Longer conversations that trigger automatic context management consume more of your usage limit.”Source: support.claude.com
Checkpoint 2 of 5· Exam question
A teacher pastes a whole term of student feedback into a single chat and keeps asking Claude to summarise themes as the conversation grows very long. What is the greatest risk to the quality of her summaries?
Correct answer: B — Detail from early in the chat may be condensed away, so her theme counts can quietly omit some feedback.
- A. Incorrect. The model in use does not change because a conversation is long. Approaching the limit affects available working memory, not model quality.
- B. Correct. As a conversation approaches the context limit, earlier messages are condensed to make room. The summary still reads confidently, but it may no longer rest on every comment she pasted in.
- C. Incorrect. Reaching a length limit does not force a plan upgrade; she can continue in a new chat or move the material into a Project.
- D. Incorrect. Condensing earlier messages does not erase them. The full chat history stays readable, so she can still go back and check.
2.When to start a new conversation
Summarisation keeps a chat alive, but sometimes a clean start is the better call. The help centre gives specific signals.
Your usage limit is running low in a long chat. A long conversation is expensive to continue, especially once summarisation kicks in. The help centre's advice is to try starting a new conversation if you're approaching your usage limit in a longer chat.
You get a length-limit error. The message reads: "Your message will exceed the length limit for this chat. Try attaching fewer or smaller files or starting a new conversation." This mostly happens in edge cases, such as a very large first message. Beyond starting over, you can break the content into smaller chunks and process them separately, summarise or extract key sections before sending, or ask Claude to find the most relevant portions first.
The chat has drifted. Because recall degrades as context grows, a conversation full of old, unrelated material can give you weaker answers even when there is still room. The table below maps each situation to the action the help centre recommends.
Keep the two kinds of limit apart when you decide. Usage limits control how much you can use Claude across all your conversations, while length limits control how long any single conversation can become. Usage limits are about quantity over time; length limits are about the depth and complexity of an individual chat.
| Situation | Recommended action |
|---|---|
| Long chat nearing the context limit, code execution on | Let automatic context management summarise earlier messages; history stays available |
| Approaching your usage limit in a longer chat | Try starting a new conversation |
| Hit a length limit | Start a new conversation or use projects |
| Hit your usage limit | Wait for the reset, upgrade your plan, or purchase usage credits |
Checkpoint 3 of 5· Check yourself
A user's very first message, with several large attachments, triggers a length-limit error in a new chat. What is the most useful next step?
The error means this one message is too large for the chat's window. Splitting it up or sending fewer or smaller files fixes it. Waiting for a usage reset fixes a different limit.
“Break your content into smaller chunks and process them separately”Source: support.claude.com
3.Persisting knowledge so a restart isn't a cold start
Starting over has a cost. Anything that lived only in the old chat is out of Claude's view, so you end up pasting the same background again. That wastes your effort and your usage. The fix is to keep durable knowledge somewhere that outlasts any one conversation.
Projects are the main tool. The best-practices article says to use projects for anything you'll reference multiple times and to upload your core working documents to project knowledge when you start. Project content is cached: when you reference it again while it's still cached, those portions count less against your limits than new content. Caches expire after a period of inactivity, though. If you come back to a project after a long break, your first message counts that content in full again. Projects also use RAG, so only the relevant parts load into a chat's context window.
Memory and past chats. On paid plans you can use Claude's memory and project summaries to build context across conversations. You can also ask Claude to search previous conversations and bring relevant information into a new chat, so you don't have to supply it again.
Keep the window lean. Tools and connectors are token-intensive, so turning off the ones a conversation doesn't need helps preserve your available context window and your usage limits. Keeping project instructions short and removing unused project files helps in the same way.
Put together, here is the decision. Restart when a chat is bloated, off-topic or eating your usage. Let it summarise when the thread itself still matters. Persist anything you'll need again, before you restart.
Checkpoint 4 of 5· Check yourself
A consultant keeps reference documents in a project and returns to it after a long break. What should they expect from their first message?
Cached project content counts less only while the cache is live. After a period of inactivity the cache expires, and the first message counts the content in full.
“If you come back to a project after a long break, your first message counts that content in full again.”Source: support.claude.com
Checkpoint 5 of 5· Exam question
A marketer runs a new campaign chat each week and re-pastes the same brand guidelines and tone-of-voice document into every one. She wants to stop repeating herself. Which sequence of steps is correct?
Correct answer: C — Create a Project; upload the guidelines as project knowledge; add project instructions; run each campaign chat inside it.
- A. Incorrect. Instructions and chats are both scoped to a project, so neither can come before the project is created.
- B. Incorrect. Knowledge is uploaded into a project, so the upload cannot precede creating one, and the chats would run without the instructions.
- C. Correct. The workspace has to exist before knowledge and instructions can be attached to it, and the chats only inherit that context once they are started inside the finished project.
- D. Incorrect. The campaign chats would run before the guidelines and instructions are in place, so the early work gets none of the benefit, and instructions cannot be added before the project exists.
Exam traps
Each one states something that sounds right. Open it to see what is actually true.
1.Letting a long chat summarise itself is free and wipes the earlier messages.Why is that wrong?
It uses more of your usage limit, needs code execution enabled, and keeps your full chat history so Claude can still reference it.
Covered in Letting a long conversation summarise itself
2.Once documents are in a project, reusing them always costs less.Why is that wrong?
The discount applies only while the cache is live. After a period of inactivity the first message counts that content in full again.
Covered in Persisting knowledge so a restart isn't a cold start
Sources
Every claim above is drawn from one of these pages, quoted as it was written on the date shown.
- 1.
“When your conversation approaches the context window limit, Claude summarizes earlier messages to continue the conversation seamlessly.”
↩︎ Letting a long conversation summarise itself“Your full chat history is preserved so Claude can reference it even after summarization.”
↩︎ Letting a long conversation summarise itself“Try starting a new conversation if you're approaching your usage limit in a longer chat.”
↩︎ When to start a new conversation“usage limits control how much you can use Claude across all your conversations, while length limits control how long any single conversation can become”
↩︎ When to start a new conversation“allows Claude to work with larger amounts of information more efficiently by only loading relevant content into the context window”
↩︎ Persisting knowledge so a restart isn't a cold start“Tools and connectors are token-intensive, so managing them helps both maximize your available context window and optimize your usage limits.”
↩︎ Persisting knowledge so a restart isn't a cold start“Longer conversations that trigger automatic context management consume more of your usage limit.”
↩︎ Checkpoint - 2.https://support.claude.com/en/articles/8606394-how-large-is-the-context-window-on-paid-claude-plansOfficial docs
“This allows conversations to continue indefinitely in most cases.”
↩︎ Letting a long conversation summarise itself“Longer conversations that trigger automatic context management use more of your usage limit.”
↩︎ Letting a long conversation summarise itself“Note: Code execution must be enabled for automatic context management to work.”
↩︎ Exam trap 1 - 3.
“Your message will exceed the length limit for this chat. Try attaching fewer or smaller files or starting a new conversation.”
↩︎ When to start a new conversation“In rare cases where you still encounter this error (such as when sending a very large first message)”
↩︎ When to start a new conversation“Break your content into smaller chunks and process them separately”
↩︎ Checkpoint - 4.
“more context isn't automatically better. As token count grows, accuracy and recall degrade, a phenomenon known as context rot.”
↩︎ When to start a new conversation - 5.
“Use projects for anything you'll reference multiple times.”
↩︎ Persisting knowledge so a restart isn't a cold start“Users with paid plans (Pro, Max, Team, and Enterprise) can use Claude's memory and project summaries to build context across conversations.”
↩︎ Persisting knowledge so a restart isn't a cold start“Giving Claude access to additional context prevents you from needing to provide the same information repeatedly.”
↩︎ Persisting knowledge so a restart isn't a cold start“When you reference that content again while it’s still cached, cached portions count less against your limits than new content.”
↩︎ Exam trap 2“If you come back to a project after a long break, your first message counts that content in full again.”
↩︎ Checkpoint