The process:
1. You put in some info about your book in the inital form.
2. A GitHub repo is created to store your book contents.
3. An outline is generated by Claude and stored in the repo.
4. The first section draft is created by Claude
5. A summary of the draft is created by DeepSeek
6. The accumulation of summaries are passed to subsequent draft Claude prompts. And the process continues.
7. After all sections have been drafted, front-matter is generated by DeepSeek
8. And the sections are all then stitched together into a single draft, and you're done
* Uses GitHub to store all drafts - connects via ENV vars
* DeepSeek and Anthropic keys also stored in ENV vars
* Uses Postgres DB to generate and store facts about respective books in a books table
* Can save progress, if the book gen process is interrupted, and continue where it left off.
* Has a retry strategy for the API calls
* Has review, and summary, and redundancy checking features
* Server logs to show you every step in the process
mackerricher•6mo ago
The process: 1. You put in some info about your book in the inital form. 2. A GitHub repo is created to store your book contents. 3. An outline is generated by Claude and stored in the repo. 4. The first section draft is created by Claude 5. A summary of the draft is created by DeepSeek 6. The accumulation of summaries are passed to subsequent draft Claude prompts. And the process continues. 7. After all sections have been drafted, front-matter is generated by DeepSeek 8. And the sections are all then stitched together into a single draft, and you're done
* Uses GitHub to store all drafts - connects via ENV vars * DeepSeek and Anthropic keys also stored in ENV vars * Uses Postgres DB to generate and store facts about respective books in a books table * Can save progress, if the book gen process is interrupted, and continue where it left off. * Has a retry strategy for the API calls * Has review, and summary, and redundancy checking features * Server logs to show you every step in the process