jev-bfs
Terminal tool that finds link paths between Wikipedia articles: Jev ranks each page's outgoing links while Python runs the search.
Overview
jev-bfs finds a path between two English Wikipedia articles, such as Chicken nugget to Heat death of the universe, and shows pages, link scores and elapsed time live in the terminal. Chromium reads the articles, Jev ranks their outgoing links by how likely they lead toward the target, and Python controls the search; there is no ChatGPT controller or search API. By default it keeps the five highest-ranked links at each level.
It installs with uv as a tool from GitHub, installs a Playwright browser, and reads the TypeSafe key from the environment, a local .env or a user config file. Output can be saved as a trace or emitted as JSON with the path, observed links, scores, progress events, errors, timing and API usage, and exit codes distinguish found, not found or error, invalid arguments and interruption. It is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How likely is this outgoing link to lead toward the target article?(inferred) | — |
How it works
- Load the start article in Chromium
- Jev scores each outgoing link against the target
- Python keeps the top five per level and expands the search
- Repeat until the target is found; report path and usage
Strengths
- Search control stays in code
- Rich JSON trace for analysis
- Clear exit codes for scripting
Caveats
- Requires a browser install and a TypeSafe key
- Results and timings vary between runs
Getting started
`uv tool install git+https://github.com/komikat/jev-bfs.git`, `jev-bfs --install-browser`, set JEVON_KEY or TYPESAFE_API_KEY, then `jev-bfs "Coffee" "Black hole"`.At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- CLI with Playwright
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Search tooling: finds link paths between English Wikipedia articles by having Jev rank each page's outgoing links while Python controls the search.
yibie/awesome-jevWikipedia link race pathfinder guided by Jev: assesses outbound links to navigate between two articles in real time.
logicrw/awesome-jev-projectsWikipedia link races with direct Jev ranking and a live terminal display.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
How promising each outgoing link is for reaching the target article.
Who controls the search?
Python, keeping the top five ranked links per level.
Can I use the output in scripts?
Yes, use --json for a full machine-readable report.
Related projects
Milvus-based multi-hop graph RAG library where two-stage Jev Noul scoring selects relations and reranks passages.
LibrarybetaPythonGraph-database-agnostic GraphRAG framework where swappable Laya or Jev decisions score edges, route intents and gate hallucinations.
ServicebetaPythonCLI and agent skill that shortlist notes in Obsidian or Logseq locally, then have Jev rerank them, returning original passages.
DemobetaJavaScript- neo4jev★ 173
Demo that navigates a Neo4j graph hop by hop, asking Jev which relationship to follow and whether the goal is reached.
LibraryprototypePython Independent experiments on Jev for search stopping, memory reranking and multi-hop relation selection, with stated limits.
Articleresearch- Jev Search★ 534
Plain-language web search where a decision model like Jev picks query, sources and time range, then ranks Search1API results.
AppproductionTypeScript