Article
CommunityJev Practices & Patterns
Jev review: tested against Fable on 520 posts
A social-scheduler founder's accuracy test of Jev: four yes/no questions over 522 captions, then the same questions for Fable on a random 100. Agreement reached 94/100 on promotion, and all 520 scored captions cost $0.011.
- Format
- Article
- Reading time
- 2 min
- Published by
- Community
- Author
- seenpaid
- Last verified
- Oct 2026
Highlights
- Jev answered four yes/no questions per caption, from spam and bait to whether a stranger would understand the post.
- 520 of 522 captions returned; 100 were re-asked of Fable with identical wording and no sight of Jev's answers.
- Agreement table: 94/100 promotion, 76/100 stranger comprehension, 74/100 spam or bait, 70/100 first line.
- The meta summary adds that Jev was the stricter judge on spam 12 to 3 and agreed 85/100 on it.
- All 520 captions cost $0.011 at $0.042 per million input tokens with free output, per the author.
Watch out
Grounded in the truncated seenpaid post and its meta description captured on 2026-10-15; the post's later sections did not make it into the digest, and Fable is by the author's own framing a baseline, not ground truth.
Questions about this entry
Short answers drawn from the linked project or page.
seenpaid is a social media scheduler that AI agents can drive; before letting Jev into its validate_post tool, which checks a caption against each platform's rules, the founder tested accuracy on 522 of his own captions.
Would a typical person see the post as spam, engagement bait or a hard sell? Does the first line give a reason to keep reading? Does it make sense to a stranger? Does it promote a brand, product, service or offer?
On a random 100 captions, the post's table shows 94/100 agreement on promoting something, 76/100 on making sense to a stranger, 74/100 on reading as spam or bait, and 70/100 on the first line.
The author says Fable is not ground truth; it is the model he would have used if Jev did not exist, which makes it the honest baseline for deciding whether Jev belongs in the scheduling check.
The meta summary says all 520 posts cost $0.011, at $0.042 per million input tokens with free output; 520 of 522 captions returned, and two failed.
The meta summary says Jev was the stricter judge on spam 12 to 3 and agreed 85/100 on it, while the visible table shows 74/100 on the spam-or-bait question; the truncated post does not reconcile the two framings.
No, and the author does not present it as one: 100 captions from one account, one baseline model and pooled judging. Use it as a workflow check before wiring Jev into a content pipeline.
More Jev resources like this
A single-file port of jev-ultrafast that lets Jev drive an Ego Lite browser on macOS: an in-page script indexes visible controls and one typed-choice request returns both the operation and the target index.
Jev Practices & Patterns#community#awesomejev#catalog-import
Mock carrier support agent with the guardrail layer built twice behind one interface: chat-model JSON judging versus TypeSafe Jev typed questions over the same 25 rules.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
deepansh-saxena/jev-guardrailsPython
deepansh-saxena/jev-guardrailsPythonjev-filter is a CLI that filters tool output before it reaches your agent and, since 0.3, drives web pages and desktop apps too. Jev picks the next operation and its target among controls the program enumerates itself; the program re-checks and executes, and irreversible steps stop for confirmation.
Jev Practices & Patterns#community#awesomejev#catalog-import
Community
apixly-ai/jev-filterPython
apixly-ai/jev-filterPythonAn 11-page working paper on placing a small commercial text judge beside a live trading lab: six tests for a text signal, what the judge may and may never touch, and the plumbing failures that corrupted its readings. No profitability claims.
Jev Practices & PatternsArticle#community#catalog-import#yibie
Community
A sparse-checkout research workspace of independent lab write-ups, two of which cover Jev: one on API behaviour, input limits and multilingual evaluation, another on moderation and pre-execution command risk checks.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
kiarina/labsDart
kiarina/labsDartUnofficial laptop study of parallel constrained decoding: prefill once, share the KV cache across fields, and decide all schema fields in one batched forward pass without writing JSON.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
rorshopping/jev-on-a-laptopPython
rorshopping/jev-on-a-laptopPythonFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.