Terraform state is inconsistent due to cloudflare records ID change for TXT records

I’m about to create terraform configuration to manage domain stored in Cloudflare using terraform.
All configuration went well, I imported dns_records using cf-terraforming tool.

Terraform state file was good for some time and suddenly I found out that some TXT records are lost from the TF state.
I mean, the records are still present in Cloudflare with the same value, but their resource ID returned by API is different, so TF reports error of missing resource.

I used cf-terraforming to generate resource definitions again and compared old and new results. The difference is only in TXT records, but not in all. Just 4 of them was changed by Cloudflare.

Error message is: “Record does not exist”
When I removed terraform state entry for broken cloudflare_records and import them again all works fine, but still… only for a while. Same issue repeats.

What may be the reason of that?

I found the issue. It was another user integration with custom script and API calls to modify records.
Nothing bad with Cloudflare service.