Your 65% stylistic noise figure is spot on, and I've seen the same thing when drafting Terraform module documentation. It kept trying to change `terraform apply -auto-approve` to a "more engaging" phrase, which completely misses the point of a precise command.
Your cost breakdown resonates. For that $30/month per seat, you could be running a dedicated linting pipeline in your CI/CD that actually understands your tech stack's vocabulary. A simple Markdown proselinter config can enforce your team's real style guide without ever questioning your `api/v1/namespace` endpoint names.
The real hidden cost is the normalization of language away from precision, like you said. When you're writing a security policy or a Kubernetes manifest explanation, clarity *is* the engagement.
Cloud cost nerd. No, I don't use Reserved Instances.
Your data on the 65% stylistic changes is a solid foundation for this discussion. It quantifies the core issue: you're paying for corrections, but mostly getting subjective rewrites.
That 'expensive style guide' label hits harder when you realize it's not even your style guide. It's pushing a generalized, non-technical voice. The moment it flags an API endpoint as a spelling error, it's shown it doesn't understand the document's purpose.
Your final point about separating functions is key. A manual proofread handles intent, a linter enforces consistency, and neither tries to make a technical command sound 'friendly'.
—AF