Video unavailable. Open the source below.
NewHow it uses Jev
The extension sends a draft plus optional reply/quote context and media descriptions as state, with the configured rubric questions in one Jev request. Rubrics use Score, Noul, or Choice types. If enabled with an OpenAI key, attached media is described before the text and description are sent to Jev.
What Jev decides
Illustrative Jev requestExample answers · not a recorded Jev response · Source ↗
Question 1 · sounds ai
YOUR APP
INSTRUCTION
Does draft read like an LLM or corporate social media manager rather than a real person?
STATE
Draft social post and optional quoted post/reply/media context.
JEV · NOUL
YesNo
App workflow
- Collect the compose state
The extension reads the draft and supported reply or quote context from the compose surface.
- Describe media when configured
An optional OpenAI vision request turns attachments or representative video frames into text descriptions; descriptions are cached per attachment.
- Ask the rubric questions
The background worker filters media-dependent rubrics and posts the shared state with all applicable typed questions to Jev.
- Show the scorecard
The content script renders rubric results and the combined verdict beneath the composer for the author to review.
Why it is interesting
Rubrics share one request and score the same contextualized draft, while media-only criteria are included only when media is present. The scorecard informs the author before posting; it does not post on the user’s behalf.