Skip to content
SEO Rapor
Log inStart freeTREN
Export and AI › Fixing with AI

Fixing with AI

Give the .md file to an AI assistant and have it make the fixes.

The to-do file (.md) is written so that an AI assistant can read it and fix the site: each item comes with a task checkbox, its reason and sample URLs. You give the file to a tool that can see your site's code (Claude Code, Cursor, ChatGPT, etc.).

#Steps

  1. On the report page, download the .md file with Export › To-do list.
  2. Put the file in your site's code folder or paste its content into the chat.
  3. Ask the AI assistant to make the fixes with a prompt like the one below.
  4. Review the changes, publish them and crawl the site again; the History tab shows what got fixed.

#Example prompt

Prompt
The attached yapilacaklar.md file came from the SEO crawl of my site.
Fix the errors in this file in order of severity and explain each change.
Before fixing, verify each item in the code; if something looks like an intentional choice, leave it alone and ask me.
Mark the items you finish with [x].

#Structure of the file

Below is a shortened example showing the layout of the file (URLs and numbers are examples).

yapilacaklar.md (example)
# SEO to-do list: example.com

- Site: example.com
- Scan date: October 18, 2026 at 2:05 PM GMT+3
- Scope: 312 addresses crawled, 4 min 12 sec

> This list comes from a live crawl. Verify in the code before fixing. If something is a deliberate choice, leave it alone and tell me.

## To fix

### Errors

- [ ] **Broken internal link** (ic-baglanti-kirik) — 1 target, on 312 pages
  - Why: The page links to an address that returns 4xx/5xx.
  - Targets:
    - https://example.com/old-page (404) ← 312 pages [menu/template: Fixed in one place: the site's menu, header or footer.]
      - https://example.com/
      - https://example.com/about
      - https://example.com/contact
      - … +309 more pages (full list in the CSV)

### Warnings

- [ ] **Meta description missing** (meta-yok) — 8 pages
  - Why: The description text in search results is picked at random from the page.
  - Example addresses:
    - https://example.com/services/seo — Meta description is missing or empty
    - https://example.com/blog/first-post — Meta description is missing or empty

## Info (not urgent)

- twitter:card missing (twitter-card-yok) — 41 pages, e.g. https://example.com/
  • Title and header: the site, the crawl date and the scope.
  • To fix: “Errors” first, then “Warnings”. Each item starts with a - [ ] task checkbox, the issue name, the check code and the number of affected pages; below it come the “Why” and sample URLs. For broken and redirecting links, the examples are grouped by target.
  • How (Claude's suggestion): if Claude's review was added to the crawl, a suggested fix is also written under the related item.
  • Site-wide: findings that are not tied to a page, such as robots.txt and the sitemap, with the measured value.
  • Info (not urgent): info-level findings in a single line.
  • At most 15 sample URLs are written for each item; the full list of the rest is in the CSV.