What exactly is sent to TypeSafe during scoring?

It sends your research profile plus titles, abstracts and categories to TypeSafe for P(True) scoring, combining overall and dimension questions per paper.

Will it miss papers if I skip several days?

No fixed window is used; the latest N papers are compared against processed IDs, so increase max_results if runs are infrequent to avoid missing older papers.

How does it recover from interrupted or failed runs?

Each success is appended to JSONL before reporting, seen IDs update after reports, and interrupted writes are rebuilt from the database on the next run.

How can I validate config without burning API calls?

Check-config validates JSON only, not keys or network; for a live smoke test set max_results to 3 and confirm reports before raising the count.

What are the default relevance thresholds?

A paper counts as relevant at 0.5 or above and high priority at 0.8 or above; both are configurable, and additional evaluation dimensions are folded into the same per-paper request.

What do the daily reports contain?

A Markdown file and a JSON file per date under daily/md and daily/json; Markdown applies the relevance filter and a display limit of 50, while JSON always contains every successful evaluation with full metadata and dimension scores.

What retry settings does it use?

The TypeSafe SDK retries up to three times after the first attempt with a one-second initial backoff and a four-second cap, a 30-second HTTP timeout, and a 60-second retry budget; arXiv uses pagination with up to three retries at a three-second interval.

CommunityJev Tools & CLIs

Paper Radar Jev

Python CLI that pulls recent arXiv papers, scores relevance with TypeSafe and writes daily Markdown and JSON rankings with resume-safe storage.

See highlights

Open github.com

Stars: 3
Language: Python
License: MIT
Published by: Community
Author: LYchoon
Last verified: Oct 2026

LYchoon/paper-radar-jev

User repository on GitHub

View on GitHub

An automated research paper radar that fetches the latest papers from arXiv, evaluates their relevance to a configurable research profile using TypeSafe AI, and ranks them by relevance score. Designed for personalized, daily literature discovery across different research domains.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

A configurable Python CLI that fetches recent arXiv papers, scores their relevance with TypeSafe AI using P(True), and generates daily Markdown and JSON rankings. Requires Python 3.11+, Git, GitHub CLI (gh), uv, and your own TypeSafe API key. Clone the project with GitHub CLI (gh). If you already have the project, run uv sync from its root directory.

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Scoring Jev for Evaluation Jev for Ranking

Used inJev for Software & DevTools

Languages: Python

Highlights

Fetches arXiv categories with configurable counts, sorting and thresholds.
Scores relevance with TypeSafe P(True) plus optional evaluation dimensions.
Writes daily Markdown and JSON reports with full metadata in JSON output.
Deduplicates by paper ID with atomic writes, locks and per-date rebuilds.

Watch out

Failed calls, empty abstracts and invalid probabilities are not marked seen and stay retryable; check logs and exit codes for gaps. 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.

More Jev resources like this

brier

Open local decision model on Qwen3-4B with LoRA adapters that answers choice, score, and noul questions with probabilities, calibrated confidence, and conformal sets, offline with no per-call cost.