DEV Community: Kelvin-Air

I made a study app that blocks the next topic until you can say it back

Kelvin-Air — Tue, 22 Sep 2026 00:41:16 +0000

I killed the video factory and kept the part of studying that actually hurts

I built Tablet because every “AI study” product I tried did the same thing: turn a lecture into more text I could ignore.

Summaries feel like progress. They are not retrieval. On exam week you still cannot say the claim out loud without looking down.

So Tablet is not a note app with a chatbot glued on. It is a retrieve loop with a binge gate.

The loop

Ingest — paste text, a URL, YouTube, or a PDF. Optional exam date.
Distill — the source is cut to essentials only, then split into short study units. Pedagogy is locked: Claim → Why → Example → Trap.
Practice unlocks immediately — Drill decks + checks as soon as distill persists. You do not wait on a render.
Retrieve — 4-option MCQ (one correct, three plausible traps), type-it, or teach-back.
Retention — same claim, colder prompt. Due times follow the exam window, or ~+2 days if you never set one.

Watch is a three-column study surface: queue, notes (markdown, generate, chat), and the next check. The title sits in the center. Side rails collapse instantly. Leftover MP4s can still play if they exist. That is not the product.

The product is: you cannot skip to the next unit until you retrieve.

Rules I will not soften

Binge gate. Next topic stays locked until a check is cleared or teach-back is logged.
Units end on an open trap. The punchline is not in the lesson. You earn it in Check or on the orb.
Teach-back is graded. Pass / shallow / confident-wrong. Confident-wrong does not clear the gate. That is the hypercorrection rule — being sure and wrong is worse than being unsure.
Mid-checks are objective. Exactly four options. No “all of the above.” acceptableAnswer is an exact match.
Long sources are split, not stuffed into one infinite chat. Aim ~15h of source; hard ceiling ~25h. Above that we refuse or force a split.
Citations from day one. Thin sources get rejected.

If that sounds strict, good. College is already timed. The app should be too.

What I deleted on purpose

I had a vertical Short encode stack. Remotion, kit-fill, SVG boards, ffmpeg compose. It was expensive, slow, and it let me pretend “shipping video” was shipping learning.

Founder overrule: that stack is gone. Practice is the face until a new encode is actually locked. Builders will understand this: the impressive demo was eating the product.

Stack (boring on purpose)

Layer	Choice
Client	Expo + Expo Router + TypeScript — one codebase for iOS, Android, web
API	Express + TypeScript
Data	PostgreSQL + Prisma
Files	Cloudflare R2
Hosting	Render
Distill	Claude primary, OpenAI backup
Voice (orb)	ElevenLabs when keyed
Billing	Stripe

No mystery framework. No “we trained our own video model.” The hard part is the contract: essentials → retrieve → space it toward the exam.

Pricing, because COGS is real

Free: 10 lessons / rolling week
Pro: $29.99 / month
Exam week: $14.99 for 7 days

I raised this on purpose. A mid-length course (~15h source) is not a 9.99 toy once you pay for distill + TTS. Fair use still applies — Pro is not “one 25h dump every billing cycle forever.”

Who it’s for

Undergrads who have a date on the calendar and a pile of PDFs they have highlighted into oblivion.

If you want a second ChatGPT window, this will annoy you. If you want something that makes you say the claim back before it lets you binge the next clip, try it.

Site: mytablet.app

Happy to talk ingest, the binge gate, or why I ripped the encoder out. Roast the loop, not the landing.

How do you use FicCrunch?

Kelvin-Air — Thu, 03 Sep 2026 09:49:51 +0000

Stop wasting time on 20 tabs staring at the screen and not getting any work done!!!!! Use FicCrunch!

Here I made a video on how you can use FicCrunch to tackle your issues in OSS,

👉 https://www.youtube.com/watch?v=n7RDrin4xdM

3..2..1 Lift off!!!

Kelvin-Air — Wed, 26 Aug 2026 02:36:33 +0000

Proud to announce that FicCrunch has launched on Peerlist! 🎉

FicCrunch turns a tangled GitHub issue/PR thread into a bounded reading order and visual graph — built to help developers stop losing hours in cross-repo link chains.

If you have a moment, an upvote on Peerlist would genuinely help: [https://peerlist.io/ficcrunchteam/project/ficcrunch]

How to beat tab chaos in Open Source

Kelvin-Air — Tue, 25 Aug 2026 14:10:34 +0000

I kept losing hours to tangled GitHub issues, so I built a tool to fix it

Every open source contributor knows this moment: you open an issue to fix a small bug, and it links to another issue, which links to a PR, which references two more issues in a completely different repo — and forty minutes later you have fourteen tabs open and you've forgotten what you were even trying to do.

That happened to me enough times that I finally built something for it.

What FicCrunch does

You paste in one GitHub issue or PR link. FicCrunch crawls the linked issues and PRs it connects to — across repos, not just within one — and gives you back two things:

A reading order — what to read first, second, third, so you're not guessing
A visual graph — the actual shape of how everything connects

The crawl is bounded, not infinite. It's not trying to summarize the entire internet about your issue — it's mapping the specific thread you're standing in.

A real example

I ran it on a currently-open Kubernetes tracking issue (kubernetes/enhancements#5526, Pod Level Resource Managers). It correctly traced the full picture across three repos — the KEP, the alpha/beta implementation PRs, and the docs — and it also surfaced something genuinely useful: why the feature got disabled in Beta, tracing back to an unresolved CPU Manager conflict still being worked on. That's the kind of context that normally takes real digging to piece together by hand.

Why structure first, AI optional

FicCrunch doesn't require AI to work. The link graph and reading order are built from the actual GitHub data — no hallucination risk, nothing to double-check. AI summaries are available as an optional layer on top, if you want them, but the core tool works without ever calling a model.

Try it

It's free to use: ficcrunch.com

I'm building this solo as a side project, so if you try it on a gnarly issue of your own, I'd genuinely love to hear how it goes — what it got right, what it missed, what you wish it did differently.

Just pumped up the theme for FicCrunch!!!

Kelvin-Air — Tue, 25 Aug 2026 14:02:47 +0000

Stop opening 20 tabs to understand one GitHub issue 🐙

Kelvin-Air — Tue, 18 Aug 2026 20:06:30 +0000

We have all been there. You find an open-source bug you want to fix, or an issue you need to triage. You click the link. Inside the discussion, someone links to a Pull Request in a dependency repository. That PR links to a closed issue. That issue points to a discussion thread.

Before you even write a single line of code, you are 20 tabs deep in browser chaos.

You lose the structural context, you forget where you started, and your brain is fried just tracking the dependency web.

I got completely sick of this tab tax, so I built FicCrunch.

🗺️ What is FicCrunch?

FicCrunch is a bounded reading path and link graph explorer for GitHub issues. Instead of forcing you to manually click through a maze of cross-repository links, it builds a structured roadmap for your reading.

[Your Target Issue] 
       │
       ├───► [Linked PR in Repo B] ───► [Fixed in Version X]
       │
       └───► [Related Issue in Repo C] ───► [Discussion Thread]

You paste a single GitHub issue URL into the platform, and FicCrunch crawls the connected threads to generate a clean, sequential reading order. It stops the tab explosion by mapping out exactly what you need to read, and in what order.

🛠️ The Core Value (And why it saves your sanity)

Bounded Reading Paths: It limits your scope. You only see the issues and PRs that actually impact the problem you are tracking.
Cross-Repository Link Mapping: If an issue spans across an organization's main app, a client library, and a backend parser, FicCrunch maps them into a single graph.
Context Preservation: You can trace a bug from a top-level alert down to the exact broken package version without losing your place.

🔒 A Note on the Engine

I am building FicCrunch as a proprietary platform, so the core graph-mapping engine remains closed-source and private.

However, as someone who spends all day in the open-source ecosystem, I wanted to build a utility that genuinely helps contributors and maintainers navigate the unglamorous, chaotic side of tracking multi-repo dependencies. You don't need to open your code to add value to the community!

💬 I'd love your feedback!

FicCrunch is live, and I want to make it as useful as possible for triage workflows.

How do you currently handle tracking issues across multiple repositories?
What features would make a tool like this an immediate part of your daily workflow?

Give it a spin at ficcrunch.com and let me know what you think in the comments!

FicCrunch has released!!!

Kelvin-Air — Tue, 11 Aug 2026 11:35:02 +0000

Having trouble starting open source on GitHub?

You pick an issue. Then the links fan out — nested threads, other repos, tabs everywhere. Motivation dies before you write a line.

𝗙𝗶𝗰𝗖𝗿𝘂𝗻𝗰𝗵 turns one pasted issue or PR into 𝗮 𝗯𝗼𝘂𝗻𝗱𝗲𝗱 𝗿𝗲𝗮𝗱𝗶𝗻𝗴 𝗽𝗮𝘁𝗵 and 𝗹𝗶𝗻𝗸 𝗴𝗿𝗮𝗽𝗵 — so you know what to open next, including when links leave the root repo.

• Reading queue in order
• Same set as a link graph
• Cross-repo links included, within clear caps

End that Loop now.

𝗪𝗲𝗯𝘀𝗶𝘁𝗲: https://www.ficcrunch.com
𝗚𝗶𝘁𝗛𝘂𝗯 𝗗𝗶𝘀𝗰𝘂𝘀𝘀𝗶𝗼𝗻𝘀: https://github.com/Kelvin-AirTMLA/ficcrunch-feedback/discussions
𝗗𝗘𝗩 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝘆: https://dev.to/kelvin_air_tmla/

Leave questions in the comments if you have any.

Hi, I'm Kelvin-Air

Kelvin-Air — Tue, 11 Aug 2026 11:23:52 +0000

I am a software engineering student who specializes in full-stack web development, machine learning and Artificial Intelligence.

I'd lovE to share my experiences and projects and always welocome to feedback and constructive critiques.

Youtube: 🌀 The Man Like Air 🌀
LinkedIn: Kelvin-Air Erayanmen
GitHub: Kelvin-AirTMLA