The writing playbook
8020.in is written with AI, openly. This playbook is the method: research, a written brief, a draft, a fact-check, a voice pass, a second-model rewrite, and a publish gate. Every article carries a pass log you can open.
Anyone can read the system. The same files are what Cursor agents read when they write or improve an article here. When a rule gets better, it gets better in public.
This playbook was created by Grok 4.6, Grok 4.7, ChatGPT-5.5, ChatGPT-5.6, and Fable 7.1, with a human overview and guidance.
Who wrote the articles
Written is the first draft. Revised means a later model rewrote or reviewed the article. Passes is how many logged passes that model ran. An article can appear under more than one model.
| Model | Written | Revised | Passes | Articles |
|---|---|---|---|---|
| ChatGPT-5.1 | 123 | 45 | 15 | 168 |
| ChatGPT-4o | 105 | 2 | 83 | 107 |
| ChatGPT-5.5 | 0 | 102 | 211 | 102 |
| ChatGPT-3.5 | 92 | 0 | 57 | 92 |
| Grok 4.7 | 0 | 70 | 156 | 70 |
| Grok 4.5 | 0 | 68 | 119 | 68 |
| ChatGPT-5.6 | 15 | 33 | 69 | 48 |
| Claude Sonnet 5 | 0 | 19 | 20 | 19 |
| ChatGPT-5 | 13 | 0 | 2 | 13 |
| Gemini 3.1 Pro | 0 | 5 | 4 | 5 |
| Fable 5 | 1 | 3 | 4 | 4 |
| Gemini 2.5 | 0 | 1 | 0 | 1 |
What you can open
- Every article ends with a "How this article was made" block: the models credited, the passes run, whether a research brief exists.
- Every pass log is at
/playbook/made/<slug>, with the brief and mechanical checks run live on the published HTML. - Every rule the writing agents follow is on this site, in the same files they read.
Start here
See the workThe processResearch -> brief -> draft -> fact-check -> voice -> rewrite -> publish.
One real articleCase studyThe brief, the pass log, and live checks for
photography.
ReusableThe engineThe brand-neutral method, and how to adapt it to another blog or a book.
For agentsCursor promptsSmall prompt packs for each writing pass.