Conclusion
Content risk checks use rules shipped with the app and run offline on the device. Results are content-expression risk references, not legal conclusions or platform-review guarantees. This feature does not upload scripts or detection content to Tdx or third-party detection servers. Its entry is shown only when the app interface language is Chinese.
Choose a detection scope
- Detection scope is single-select, so each check uses one scope.
- All rules runs every enabled rule and is useful when the publishing scene is not yet known.
- News media, Advertising, or E-commerce runs the common rules plus rules that apply to that scene.
- General promotion, Sensitive content, News wording, Beauty & health, E-commerce, and Social media are six result filters. They do not rescan the content and are not multi-select detection scopes.
Start from the editor
- Tap the content risk check icon at the bottom of the script editor.
- The app saves the current script and opens the preparation page. Confirm the script and choose one detection scope.
- Tap Start check, wait for the complete result, and review, replace, or ignore each item.
- Free text edits make the old result stale. Tap Check again to generate a result for the current text.
Cues and ordinary brackets
Only static cues created with Insert cue are excluded from checked text. Bracketed text such as [note] that you type or paste normally remains ordinary script text and is checked; the editor does not require brackets to be paired.
Local overrides
Ignore in this script affects only the current script. The global whitelist and custom risk terms affect every script on this device. They and temporary detection drafts stay in private app data and are not included in ordinary analytics events.