# How Jeff Geerling actually uses coding agents

Ansible, Raspberry Pi YouTube

**If you take one thing:** `AI Disclosure`, README names the exact model. every Claude-built repo we found ends with a section saying which model wrote what, down to the effort level.

- **Agent:** Claude (web) ([source, 2026-09-23](https://github.com/geerlingguy/timemachines-flash/blob/d70e42a78989f5b547f40cc9b0cd639401fd3087/README.md#L68))
- **Model:** Claude Opus 5.5 ([source, 2026-09-23](https://github.com/geerlingguy/timemachines-flash/blob/d70e42a78989f5b547f40cc9b0cd639401fd3087/README.md#L18))
- **Model:** Claude Fable 5.1 ([source, 2026-09-08](https://github.com/geerlingguy/BigClock-Classic/blob/1e111553dd5f0295c588a5451b08bda0cef070b5/README.md#L32))
- **Review:** Reviews in stages, like a tech lead with juniors ([source, 2026-01-20](https://www.jeffgeerling.com/blog/2026/migrating-13000-comments-from-drupal-to-hugo/))
- **Harness:** Claude chat: one prompt plus attached files, then iterate ([source, 2026-09-06](https://github.com/geerlingguy/vcfmw21-map/blob/a30e78525139ef52d2d919bf02a3426ba57f47ec/README.md#L7))

## In their words

> The README is entirely my own work, but the Python script was written by Claude Opus 5.5.

> Opus 5.5 took about 8 minutes and spat out `tmflash.py`.

> I built this using the assistance of Claude Fable 5.1, since I've never built a Classic Mac OS application before.

> I'd label myself an 'AI skeptic', but I admit it's impressive how well LLMs achieve certain tasks, especially if you treat them like junior devs on a small team, break down work into reasonable-sized tasks, review the work in stages (checking in code in a VCS)

> I ran the table plan and Exhibitor CSV file through Claude Opus with the instruction:


Archetype (our label): Skeptic. Newest source: 2026-09-23.
Card: https://www.prehistoriq.com/stacks/geerlingguy/ · Data: https://github.com/prehistoriq/stacks/blob/main/data/stacks/geerlingguy.json