Skip to content
Softcyb.org
Join
Follow this build — @mina
← LoadoutStepsEquipmentAll builds
Mina Choi avatar

Build guide

Mina Choi's Research Ranger build

Work through the steps in order. Tick each one off as you go; progress is saved in this browser.

Build guide · updated Sep 6, 2026

A literature pipeline: nightly arXiv digest, adversarial summaries on demand, and a citation database you can actually trust.

Easy
Steps
5 (1 optional)
Time
1 h 50 min
One-time
$1,398
Monthly
$20/mo

Walkthrough

Progress: 0 / 5 steps (0%)1 h 50 min left
Next: Step 1Saved in this browser
  1. Step 1 of 5 · next

    Set up the literature database

    A Notion database with columns: paper, why it matters, the model's objection, my verdict, and 'why I believe this' for every citation.

    Easy20 minFree/mo (Notion education plan)Equips: Notes & tasks

    Cost note: Notion education plan

    You need
    • Notion account
    Links
    Site:Notion for education (opens in a new tab)Tip:Citation hygiene
  2. Step 2 of 5

    Schedule the arXiv digest

    A cron job at 02:00 fetches new listings for your keywords, asks the model to rank them, and writes the top five into the database.

    Medium45 min$20/mo (Gemini via Google One AI Premium)Equips: Automation

    Cost note: Gemini via Google One AI Premium

    You need
    • Step 1
    • A machine that is on at 02:00 (laptop on a charger counts)
    • Python 3
    Links
    Docs:arXiv API user manual (opens in a new tab)Docs:Notion API (opens in a new tab)Tip:Nightly arXiv digest

    crontab -e

    0 2 * * * ~/bin/arxiv-digest.sh >> ~/logs/arxiv.log 2>&1
  3. Step 3 of 5

    Make the adversarial-summary prompt

    One saved prompt: 'What is the weakest part of this paper and what would a hostile reviewer say?' Use it on every PDF before reading.

    Easy10 minFreeEquips: Main model
    You need
    • Step 2
    Links
    Tip:Argue against the paper
  4. Step 4 of 5

    Transcribe seminars locally

    whisper.cpp on the laptop. Record on the phone, drop the file in a folder, get a transcript next to your notes.

    Medium30 minFreeEquips: Voice
    You need
    • Homebrew or a C compiler
    Links
    GitHub:ggml-org/whisper.cpp (opens in a new tab)

    Run

    $ brew install whisper-cpp
$ whisper-cli -m models/ggml-base.en.bin -f seminar.wav -otxt
  5. Step 5 of 5 · optional

    Read in timed sprints

    Twenty-five minutes on the Watch timer, one paper, then a two-minute spoken summary that goes into the database unedited.

    Easy5 min$399 (Any Apple Watch; a kitchen timer also works)Equips: Wearable

    Cost note: Any Apple Watch; a kitchen timer also works

    You need
    • Step 1
    Links
    Buy:Apple Watch (opens in a new tab)Tip:Reading sprints with a wrist timer
Final equipment

What the character sheet looks like when every step is done.

Back to loadoutAll builds
Community build5 steps