01
It starts by itself
every 21 days. Runs in Claude Code routine or GitHub Actions.
- ↻ every 21 days
0 6 */21 * *
Every 21 days the loop reads its task board, pulls search data, writes one page per finding, lets a second agent grade it, and opens a pull request. It runs by itself.
Read https://gethabit.dev/loops/seo-loop.md and install this loop in my project. Show me the files first.
469 → 941 → 1,688 visibility over 3 weeks · Search Console, one site
When does it start? How does it check its work? What does it remember?
Stops after 3 tasks per run.
Read open tasks and pages that are due for their 21-day check.
Pull search data and see who ranks for each target term.
One task per finding. Skip what the board already has.
If the page exists or the task is a duplicate, stop here.
Write the page into the repo with title, description and links.
A second agent with fresh context grades it. Below 17 of 20, rewrite once.
Open a pull request. A human merges.
Read Search Console. Ranked but no clicks? The task reopens.
It writes only in its own folder and asks you before anything goes out.
A cron job starts something. This loop also checks its own work twice: once before shipping, once 21 days later with real search data.
No. It opens a pull request. You merge.
Yes. The prompt above installs from their original repos and sets up the habit around them.
Yes. The skills are open source and Habit is free. The habit runs where your agent already runs.
Read https://gethabit.dev/loops/seo-loop.md and install this loop in my project. Show me the files first.