Overview & result
eijiaraki shows toxic-filter, built on Jev: Chosen expressions on X get blurred out.
From the original post
toxic-filter by @eijiaraki: Chosen expressions on X get blurred out.
How Jev fits in the loop
- The Chrome extension classifies posts on X with Jev.
- You choose which of six categories to hide (discrimination, provocation, hatred or contempt, calls to attack, cynicism, resignation) and the threshold.
- Matching posts are covered with a translucent overlay.
- A button on the overlay shows the post again.
How to reproduce
- Review the source documentation and repository at https://github.com/eijiaraki/toxic-filter.
- Configure environment with TypeSafe Jev / System One API keys.
- Run test prompts against the decision endpoints.
Why this build matters
Shows Jev in practical use for content, marketing and SEO.
Limitations
- The author states that judgments can be wrong and do not assess whether a post is correct or the poster's character.
- You need your own Jev API token, and usage is billed to your TypeSafe account.
- The interface is in Japanese and requires Chrome 120 or later.