No API keys or settings are needed. After you install the plugin, its actions appear under Plugins in the workflow editor.
FIRST AUDIT IN FIVE MINUTES
1. On any page, add an Input and a Button.
2. In the button's workflow, add the action "GEO Audit: audit a page". Set "Page URL" to the Input's value. Leave "Custom user agent (optional)" empty and keep "Cross-check sitemap against robots.txt" checked.
3. In the next step, save the results you want, for example score, summary, fixes and report_json, to custom states or to a record.
4. Preview the page. Type the full address of a public page, starting with https://, and click the button. A run takes a few seconds.
5. If body_words is 0, a crawler that skips JavaScript gets no body text from that page. Work through the fixes list.

ADD SCHEMA TO A DYNAMIC PAGE
6. On the data type the page shows, for example Blog Post, add a text field named geo_header.
7. In the workflow that creates or updates the record, add the schema action that fits: "GEO Schema: Article", "GEO Schema: FAQ", "GEO Schema: Product or App", "GEO Schema: Business" or "GEO Schema: Breadcrumbs". For an article, fill in "Headline", "Page URL", "Author name", "Date published" and any other fields you have.
8. In a later step, save the result's header_html to the record's geo_header field.
9. Open the page that shows the record. Double-click the page background to open the page property editor. In the "Page HTML Header" box, insert dynamic data: Current page Blog Post's geo_header.
10. To use more than one builder on a page, save their header_html results one after the other in the same field.
11. Check that your privacy rules let logged-out visitors see geo_header. Then audit the page again. The types you added should appear in schema_types.

ADD A CRAWLER SNAPSHOT
12. In the same workflow, add "GEO Snapshot: build crawler snapshot". Set "Title" and "Body text" from the record. Choose the "Body format" that matches how the text is stored.
13. Optional: fill in "Short summary (optional)". Give "Link labels (optional)" and "Link URLs (optional)" as two lists in the same order.
14. Keep "Wrapper" on noscript unless you have a reason to change it. Use the same text that visitors see on the page.
15. Add the result's header_html to geo_header, after the schema. Audit the page again and check snapshot_words.

BUILD ROBOTS.TXT
16. In a workflow on an admin page, run "GEO Robots: build robots.txt". Set "Allow AI search bots", "Allow AI assistants fetching for a user" and "Allow AI training bots". Fill in "Private paths (/admin, /dashboard)" and "Sitemap URLs". Keep "This is a Bubble app (block /version-test/)" checked.
17. Save the result's robots_txt to a custom state and show it in a multiline input. Read the notes list too.
18. Copy the text. Go to Settings, SEO / metatags, "Customize robots.txt", paste it and save. The plugin cannot change this setting for you.

BUILD LLMS.TXT
19. Run "GEO llms.txt: build llms.txt" the same way. Set "Site name" and "One-paragraph summary". Give "Page titles" and "Page URLs" as lists in the same order. "Page descriptions (optional)" and "Section title (default Pages)" can stay empty.
20. Copy the result's llms_txt into a plain text file named llms.txt. Upload it under Settings, SEO / metatags, "Hosting files in the root directory".

SHOW THE REPORT ELEMENT
21. Draw the "GEO Audit Report" element on the page from step 1.
22. Set "Report JSON (from the audit action)" to the place where you saved report_json in step 3.
23. Choose an "Accent color" and check or uncheck "Show the list of fixes". The scorecard redraws when the saved report changes.

NOTES
- The Page HTML Header and the SEO settings may need a paid Bubble plan.
- The audit reads public pages only, one page per run, as a logged-out visitor. It uses workload like any server action.
- Every action returns an error value. It is empty when the action worked.
- Support: Hello@redcenterstudio.com