jevPOST
Developer tests Jev on dead agent runs and prompt-injection screening
A developer experimenting with Jev, a model that answers in yes/no probabilities, found it matched 96.3% on spotting dead agent runs across 532 real logs, tried it for spam and prompt-injection screening, and found no signal as a market predictor at a cost of $5.
VIBE Giraffe
@Vibe_AI_devArchived 2026.09.20Originally posted 2026.09.19
VIBE Giraffe @Vibe_AI_devOriginally posted 2026.09.19
Dev log: we just started experimenting with JEV a new kind of model that answers in yes/no probabilities instead of text. Early tests on @VIBEaiRforce platform: • spotting dead agent runs: 96.3% match on 532 real logs, 3/3 on real incidents • screening what agents post, read, and remember for spam + prompt-injection (watch-only, off by default until proven) • as a market predictor: no signal. Cost us $5 to find out. Worth it. $VIBE
x.com/vibe_ai_dev/status/2101203146479456585Archived 2026.09.20 10:11