junshi-research

junshi-research / research-junshi

Public

A Claude Code skill that acts as your daily ๅ†›ๅธˆ (strategic research advisor).

36
1
100% credibility
Found Mar 21, 2026 at 36 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

Junshi is a Claude Code skill that analyzes a researcher's papers to create a profile, monitors new literature from arXiv and key venues, and generates ranked, actionable research ideas with experiments and risks.

How It Works

1
๐Ÿ“š Discover Junshi

You hear about Junshi, a helpful assistant that reads your research papers and suggests fresh ideas tailored just for you.

2
๐Ÿ› ๏ธ Add it to your workspace

You simply download and add Junshi to your AI tools, making it ready to use right away.

3
๐Ÿ’ฌ Share your research world

You tell Junshi where your papers are and describe what you're working on, like handling tricky data problems.

4
๐Ÿ” Receive your first ideas

Junshi scans your papers, checks the latest research news, and delivers a ranked list of bold new directions with easy first experiments.

5
Choose your routine
๐Ÿ—ฃ๏ธ
Ask on demand

Just chat with Junshi anytime for the latest digest.

๐Ÿค–
Set it automatic

Turn on morning deliveries so ideas arrive without lifting a finger.

๐ŸŽ‰ Unlock new discoveries

Every day, you get personalized research sparks that connect your work to cutting-edge gaps, complete with experiments and risks to test.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 36 to 36 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is research-junshi?

Research-junshi is a Shell-based Claude Code skill that acts as your daily strategic research advisor, reading your papers to build a personal profile, scanning arXiv and key venues for new work, and delivering 3-5 ranked ideas with pitches, first experiments, and risks. It solves the overload of generic literature alerts by tailoring digests to your methods and gaps, saved automatically to a local folder. Install it free via git clone into your Claude Code skills directory, reload plugins, and run simple CLI prompts like "Run research-junshi" with your papers path.

Why is it gaining traction?

It stands out from basic RSS feeds or generic AI summaries by grounding ideas in your own papers first, pushing bold cross-pollinations over safe recaps, and offering cron automation for headless daily runs via the Claude Code CLIโ€”no session needed. Developers notice the ranked outputs with feasibility scores and one-afternoon experiments, plus easy profile updates and venue customization for fields like ML or economics. The claude code skills integration feels native, with clear docs covering install on macOS/Linux.

Who should use this?

Researchers in machine learning, econometrics, biology, or robotics who juggle papers in ~/papers/ and track venues like NeurIPS, AER, or ICRA. PhD students or profs wanting morning digests of arXiv hits filtered to their high-dimensional confounders or causal inference focus. Anyone using Claude Code CLI for claude code skills and needing an advisor that proposes testable directions daily.

Verdict

Try it if you're in Claude Codeโ€”free claude code install and strong user-facing docs make the 36 stars and 1.0% credibility score forgivable for an early-stage skill. Low maturity means watch for automation quirks like the skip-permissions flag, but it delivers real value for personalized research tracking out of the box.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.