Stack Exchange API write endpoint testing boundary without OAuth sandbox

Context

The Stack Exchange API provides unauthenticated read access with a modest daily quota and a client key that raises that quota, but write operations require an OAuth 2.0 access token with an explicit scope. There is no documented sandbox environment that accepts write calls without a real user account and a live token.

Goal

Determine the exact permission boundary for exercising write endpoints (e.g., /posts/{id}/comments/add, /answers/add) during integration development without using production credentials or creating public content on the live site.

Constraints and uncertainty

• Client keys are not secrets and do not authorize writes; the OAuth token is the only credential that crosses the write boundary.
• Some write endpoints mention a "preview" parameter, but documentation does not clarify whether preview mode persists content or merely validates the request shape.
• Rate-limit and backoff behavior can be observed anonymously, yet the write path remains gated behind a token that must be obtained from a real user flow.

Questions

1. Does the preview=true parameter on write endpoints suppress side effects entirely, or does it still create visible content under the authorizing account?
2. Is there any supported mechanism—such as a dedicated test site, a mock token issuer, or a documented "dry-run" mode—that allows an integration to exercise the full write request/response cycle without a production OAuth token?
3. If no sandbox exists, what is the recommended pattern for CI/CD pipelines to validate write payload construction and error handling without posting to the live network?