Long-form
Outline first, draft one section at a time, and keep voice and decisions consistent with project instructions and a short log.
-
How Do You Write a Long Document With Claude? Work from an approved outline and have Claude draft one section at a time, reviewing each before it moves on. Level 2: Context
- 1Approved outline
- 2Draft one section
- 3Edit and approve
- 4Next section
Then repeat -
How Do You Keep a Long Claude Draft Consistent Across Chats? Keep a short decision log listing the claims made, terms defined, figures used and choices settled so far. Level 3: WorkflowsDecision log
- Claims made
- Terms defined
- Figures used
- Choices settled
-
How Do You Make Claude Keep the Same Voice in Every Chat? Write a short voice guide with sample sentences and put it in your project's instructions, so you don't have to paste it into every chat. Level 3: Workflows
- Instructions for Claudeevery conversation
- Project instructionsonly inside that project
- A skillwhen activated, across projects
-
How Do You Get Claude to Outline Before It Writes? Tell Claude to write an outline and to stop there. Level 2: Context
- 1Ask for an outline, then stop
- 2You edit the outline
- 3You approve it
- 4Draft to that version
Size isn't the main problem
Anthropic lists context windows of 1M or 500K tokens for most current models in chat, so Claude can hold a long draft. On paid plans with code execution on, it also summarizes earlier messages as a conversation nears the limit, and keeps your full history. What goes wrong in long pieces is usually a wrong turn found late, or a decision that a new chat doesn't know about.
Three habits and where they live
Approve an outline before any prose exists. Draft one section at a time, ideally in a Claude Docs document you can type in and comment on. Keep the voice guide in project instructions and a decision log in **project knowledge**, because Anthropic says chats in a project don't share context unless it's in the knowledge base.
Can Claude write a book?
It can draft long manuscripts in stages, but the quality depends on your outline, sources and review. Work chapter by chapter in a project, with your voice guide in the instructions and a log of decisions in project knowledge. Check all facts yourself.
Why does Claude forget things in a long project?
Chats in the same project don't share context, unless the information is in the project knowledge base. Save important decisions as a file there. In a single very long chat, Claude may also summarize earlier messages, so an explicit log is safer.
What happens when a Claude chat gets too long?
On paid plans with code execution on, Anthropic says Claude summarizes earlier messages to make room, so a chat can often continue. Your full history is preserved, but conversations that trigger this use more of your limit. Without code execution on, this doesn't happen.
Should I use Claude Docs or a normal chat for a long document?
Use Claude Docs if you want to edit in place, leave comments or collaborate, since it's a shared document saved in your Artifacts tab. Docs count toward your usage limits. Export a copy of anything important, because Anthropic says deletion is permanent.