I encountered an error or it got stuck
Troubleshooting steps for installation startup, model connection, generation pause, task records, and Knowledge Base retrieval
It's my first time using it
Follow the creation wizard from configuring the model to getting Chapter 1 drafted
Installation and Data
Where is data stored, and how to back it up
Where is data stored, and how to back it up
See Installation & Download for “Where data is saved”. Before backing up, completely exit the application, then copy the entire data directory.
Why can't the portable and Mac versions auto-update
Why can't the portable and Mac versions auto-update
Only the Windows installer version supports in-app updates. For the portable version, download the new version and replace the program while keeping the
-data folder; for the Mac version, download the new .dmg and overwrite the installation, data is unaffected.Do I need to prepare my own model
Do I need to prepare my own model
Yes. The workbench does not include built-in model quotas; it uses your own API configuration, or you can connect a local model. The quick setup will pop up on first open; see Model Configuration.
Creation Process
Why does fully automatic also stop
Why does fully automatic also stop
Fully automatic only executes the scope you authorized. For example, if you submit Chapters 1 to 3, it stops after writing them and will not write Chapter 4 on its own; you need to select the next segment.Additionally, it will pause when encountering model unavailability, needing to re-plan, reaching the usage protection limit, or data integrity issues. Pause does not equal failure; progress and saved body text are preserved. How to continue, see Troubleshooting.
There are quality reminders on chapters, should I stop and fix them
There are quality reminders on chapters, should I stop and fix them
The default “prioritize completing the whole book” strategy records ordinary local issues as chapter reminders and continues writing. To be stricter, you can switch to “quality first” in Settings → Auto Director (设置 → 自动导演), or set it separately for a particular book.
Which should I choose, Director V1 or V2
Which should I choose, Director V1 or V2
V2 is recommended for new books. Existing novels can continue using the original version; before switching, you need to first end current creation. Switching only saves the selection and will not automatically start generation. Comparison, see Quick Start.
What is the single-chapter usage limit
What is the single-chapter usage limit
The V2 single-chapter default limit is 100,000 Tokens, accumulating generation, checking, text repair, and material organization calls for that chapter, used to prevent single-chapter consumption from spiraling out of control. Reaching the limit will pause and preserve progress; you can add more quota to continue. See AI Usage for details.
Knowledge Base
Can I use it without configuring Qdrant
Can I use it without configuring Qdrant
Yes. Qdrant is only used for Knowledge Base vector retrieval; it is not a prerequisite for starting a book and writing Chapter 1. To first run through the main line, set
RAG_ENABLED=false when running from source.Reporting Issues
When submitting to GitHub Issues, try to include:- Application version (version number in the top bar) and system
- Operation entry point and steps
- Model provider and model name used
- Pause reason or original error text, screenshots
- Desktop version log package (saved in “Version & Updates”)