English Β· νκ΅μ΄
m1kapp/fixearly Β· fixearly.m1k.app
A JS/TS analyzer that scores the cost of changing this code later β not style, not coverage. Fully local, no upload, no config:
npx fixearlyThe point isn't the grade. It's that the grade has to be falsifiable. So the ruler gets tested against real code β the 74 open-source repos in the banner are the baseline β and a scoring axis only gets in if it clears one bar:
Note
An axis has to move the score and produce a pull request a maintainer actually merges. Four candidate axes passed the statistics and were cut anyway, because nothing shippable came out of them.
Found by the tool, verified by hand, benchmarked before sending.
| Merged | What |
|---|---|
| nocodb#14309 Β· 64.4kβ | user field validation O(nΒ²) |
| outline#13117 Β· 39.9kβ | markdown import merge O(nΒ²) |
Four more approved and waiting to merge β novu, medusa, n8n, vite. Four were closed.
The full log β wins and losses β is IMPACT.md. It's generated from the GitHub API, so the ones that didn't land can't be quietly dropped.
Checkable, not claimed β the contributor graphs of nocodb and outline list me.
π§° @m1kapp β the workshop
Everything I ship lives in one org, mapped at m1k.app. Small tools, each one built because I wanted it to exist. All of these are live right now.
| Live | What it does | |
|---|---|---|
| fixearly | fixearly.m1k.app | Scores JS/TS by change cost, benchmarked on 74 OSS repos |
| ν«μ»· / ytcc (private) | ytcc-next.vercel.app | Finds a video's best moment from its own comments |
| Claude Run | claude-rank | Claude subscription value leaderboard, from submitted usage reports |
| PromptWing | promptwing.vercel.app | AI image prompt studio |
| logodown | logodown.vercel.app | Make logos the way you write markdown |
| kit | kit-inky.vercel.app | The shared UI kit the rest of these are built from |
| m1kskills | β | Markdown prompt-skills, portable to any AI tool |
Also: μ€μμλ κ³μ°κΈ° (Korean income calculator) Β· formychildren (private) β a birthday game for my kid, and the only thing here with a real user.
Contributor to nuxt-seo. Earlier and now retired: teamlog (team blogging) Β· cutin (basketball video editing). I write at uminoh.tistory.com.
The banner isn't decoration β it's the live grade distribution of the fixearly corpus, regenerated
by tools/make-banner.py. The β² marks where fixearly
scores itself.