What it is
A Python CLI that finds a link path between two English Wikipedia articles. Chromium fetches pages, Jev ranks outgoing links, and Python controls a beam search across BFS levels. For people exploring Wikipedia connections.
How it uses Jev
At each level, Jev receives the target query plus candidate link titles and URLs, and scores up to 128 candidates per request to favor links with the fewest expected remaining hops. Python keeps the top-ranked links (default width 5) for the next level. Article text is not sent to Jev.
Primitives:score
Technique worth stealing
Use Jev as a cheap per-level link ranker inside a Python-controlled beam search, keeping the search loop outside the model.
Try it
uv tool install git+https://github.com/komikat/jev-bfs.git, run jev-bfs --install-browser, set JEVON_KEY, then jev-bfs "Coffee" "Black hole".
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.90
- Shows a System One patternunclear0.50
- Handles uncertaintyno0.04
- Measuredno0.13
- Runnableyes0.97
- Worth recommendingunclear0.43
- Model replicano0.05
- Problem scopescore on a 0–2 scale0.06
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.