GraphPaper
Local-first Windows writing studio with knowledge graphs and author voice, where optional Jev evaluates candidates and editorial decisions.
Overview
GraphPaper is a local-first writing studio for Windows covering nonfiction, polemic, fiction and scientific manuscripts. You upload source material, explore it as a knowledge graph in a spatial graph studio, choose an angle, shape an outline and refine a draft, guided by your brief and a saved voice. It supports Codex or ChatGPT sign-in or your own model APIs, with separately configurable writer, editor and extraction models, plus reusable voice graphs and humanising or de-slopping tools.
Jev plays an optional supporting role: it evaluates graph candidates, evidence gaps and revision decisions, receiving compact graph context with automatic request sizing rather than the full source pile. The writing itself is done by the configured language models. Releases are distributed as a Windows download, and Graphify graph extraction supports Codex sign-in.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which graph candidate or angle is strongest for this brief?(inferred) | — |
| Yes / no | Is there an evidence gap here?(inferred) | yes · no |
How it works
- Upload sources and build the knowledge graph
- Optionally Jev evaluates candidate angles and gaps
- Outline and draft with the writer model in your voice
- Optionally Jev judges revision decisions
Strengths
- Local-first with user control of argument and voice
- Jev optional and fed compact context
Caveats
- Windows only
- Jev role is optional and secondary
Getting started
Download the Windows release and follow the quick start; sign in with Codex/ChatGPT or add your own model API.At a glance
- Type
- Project
- Status
- beta
- Language
- TypeScript
- Runtime
- Windows desktop app
- Has code
- yes
Use cases and integrations
How curated lists describe it
A Windows writing studio with knowledge graphs, author voice, humanizer/deslopping, project folders and Codex/ChatGPT sign-in.
hellogumbo/awesome-jev
Frequently asked questions
Does Jev write text?
No, it optionally evaluates candidates, evidence gaps and revisions.
Which models write?
Codex/ChatGPT via sign-in or your own configured models.
Which platform?
Windows.
Related projects
Linter for Obsidian vaults and LLM wikis that flags contradictions, stale claims, unresolved markers and dangling links with Jev.
ProjectbetaPythonCodex plugin that runs polls on simulated reader profiles with a System One model, recording typed answers for follow-up studies.
ProjectbetaTypeScriptPi extension where Jev decides whether a reply needs rewriting, then your model clarifies it and may add a small visual.
PluginbetaTypeScriptWindows file organiser that uses Jev Choice and Noul to suggest which folder a new file belongs in and to drill down to find files.
ProjectprototypeTypeScript- neo4jev★ 173
Demo that navigates a Neo4j graph hop by hop, asking Jev which relationship to follow and whether the goal is reached.
LibraryprototypePython Chrome and Firefox extension that hides AI-written posts on X, LinkedIn, Reddit and YouTube by asking Jev 30 narrow style questions.
Browser extensionproductionJavaScript