User repository on GitHub
Experimental: live X draft viral scorer powered by TypeSafe Jev
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Tools & CLIs
Draftpulse
A local React and Express editor that scores an X draft across six dimensions, from hook strength to a slop penalty, by asking Jev several narrow rubric questions in one structured call, with a deterministic fallback when no key is set.
- Stars
- 1
- Language
- TypeScript
- Published by
- Community
- Author
- pekth
- Last verified
- Oct 2026
About this project
Live viral-dimension scoring and calibration engine for X posts before publishing. DraftPulse provides real-time viral-dimension scoring and category classification for social drafts as you type, highlighting structural strengths and AI slop risks before you hit publish. Live Debounced Scoring: Automatically evaluates drafts on typing pauses (500ms), giving instant visual feedback without blocking the editor.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
- Languages
- TypeScript
Highlights
- Live scoring debounces roughly 500 ms after a typing pause instead of firing on every keystroke.
- Six dimensions: hook strength, specificity, reply propensity, shareability, dwell time and a slop penalty.
- TYPESAFE_API_KEY stays in the Express backend, and unsent drafts go only to the local scoring proxy.
- The genre is auto-detected (Build in Public, Technical Insight, Hot Take, Story, Question) to pick the rubric.
Watch out
The project calls itself an experimental research prototype whose heuristic scores do not guarantee reach or algorithmic promotion, and nothing here contradicts that. No accuracy figures, correlation against real engagement, or before-and-after examples are published, so there is no evidence the scores track anything measurable. The composite weighting is configurable through server/weights.ts, but the README does not say how those weights were derived or what corpus they were fitted on, and the calibration loop is described as a plan in docs/data-loop.md rather than a shipped pipeline. The zero-scraping policy rules in the official X API, owned posts and licensed archives without stating which credentials or API tier that needs, so the calibration path is untested in practice. Only one check command is offered, npm run test:compose, with no coverage numbers. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
Closer to a structure checker than a predictor. The dimensions are concrete: the first 80 characters for hook strength, concrete numbers and exact tools for specificity, insight density per character for dwell time, and a negative penalty for cliches, em-dashes and boilerplate phrases. The README is explicit that heuristic scores give structural feedback and do not guarantee platform reach or algorithmic promotion.
Yes. A deterministic fallback built on heuristic analysis covers the scoring path, so npm install, cp .env.example .env and npm run dev is enough and TYPESAFE_API_KEY is optional for mock mode. Two processes have to be running: the UI on port 5174 and the scoring API on 8787. With a key set, Jev evaluates several narrow rubric questions in a single structured call.
Composite scoring combines the per-dimension ratings through configurable weights in server/weights.ts, and the repo points at a data-loop plan in docs/data-loop.md for refining the heuristics against observed post performance. What is not documented is how the weights were originally fitted, on what data, or how much signal the scores actually carry. The dimension matrix lists weight impact qualitatively, high, medium or negative, rather than numerically.
According to the README, no. Unsent drafts stay in local application state and are transmitted only to the local scoring proxy endpoint, and the API key lives in the Express service rather than being bundled into the client. The stated calibration policy is zero scraping: it relies on the official X API, owned posts or licensed archives, and does not bypass platform access controls.
It needs Node.js 18 or newer. Run npm install, copy .env.example to .env, then npm run dev. The UI launches on http://localhost:5174 with the scoring API on http://localhost:8787.
Hook strength, concrete specificity, reply propensity, bookmark shareability, dwell time, and a slop penalty. Weights live in server/weights.ts, and the composite formula combines the dimensions into an overall readiness score you can configure.
Drafts are scored on typing pauses, roughly every 500 ms, so feedback appears without blocking the editor. The server also auto-detects the post genre, such as Build in Public, Technical Insight, Hot Take, Story, or Question, to pick the right rubric.
More Jev resources like this
4GitHub stars
Shell CLI that sends state plus noul, choice, and score questions to Jev in one request and prints probabilities, choices, and rubric positions for pipes and scripts.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
y0usaf/typesafe-cliTypeScript
y0usaf/typesafe-cliTypeScript38GitHub stars
Python memory filter that judges every stored memory against the current request with one calibrated yes/no per memory in a single Jev request, keeping everything above a threshold.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
samdotmak/jev-recallTypeScript
samdotmak/jev-recallTypeScript27GitHub stars
Pi extension that replaces many small tool calls with one typed TypeScript execute_tools call, type-checked before running in a sandbox.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
boozedog/pi-codemodeTypeScript
boozedog/pi-codemodeTypeScript30GitHub stars
Forkable weekday radar that asks Jev one yes-or-no question per interest over each new arXiv announcement and publishes a page plus RSS feed.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
Eliot5566/JEV-Paper-RadarPython
Eliot5566/JEV-Paper-RadarPython17GitHub stars
Pi extension that trims wasted context before each model call by deduplicating repeated tool results, stripping failed payload args, and letting the model summarize closed spans.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
Davidcreador/pi-dcpTypeScript
Davidcreador/pi-dcpTypeScript38GitHub stars
Async supervisory layer for Hermes Agent that locks Definitions of Done, proves mechanical criteria locally, and calls Jev only for unresolved semantic checks under a token budget.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
keeltrace/hermes-nervePython
keeltrace/hermes-nervePythonFrom the Jev guides
Original write-ups on what this entry is for.
The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.
Jev Tools & CLIs#gateway#vercel#cloudflare
Read article
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.
Jev Tools & CLIs#classification#email#security
Read article
From the Jev community
Posts from builders shipping with Jev right now.