DEV Community

SweepBase
SweepBase

Posted on

How I Maintained an Awesome-List of 136 Crypto Cards as a CI-Linted Dataset

#opensource #github #webdev #showdev

Last month I open-sourced awesome-crypto-cards — a curated list of 136 crypto debit and credit cards. This post is about the boring infrastructure: why I run awesome-lint in CI, how I keep the list synced with the dataset behind sweepbase.net, and where I underestimated effort.

Why a flat README, not a database

The list lives as a single README.md. No JSON, no YAML, no static site. People who land on a GitHub awesome-list expect to scan markdown, not click into an interactive viewer.

Trade-offs I accepted: no programmatic queries, no filtering UI, no auto-generated content.

Trade-offs I avoided: an extra build step, broken links from generator bugs, and the friction of "wait, where do I edit this?"

The awesome-lint CI

Every push runs awesome-lint via GitHub Actions. It catches:

  • Duplicate URLs (you'd be surprised)
  • Links missing https://
  • Markdown formatting that breaks GitHub's renderer
  • Broken anchor references in the contents section
  • Categories that don't sort alphabetically 
# .github/workflows/main.yml
name: Awesome Lint
on: [push, pull_request]
jobs:
  lint:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with: { node-version: '22' }
      - run: npm ci
      - run: npx awesome-lint
Enter fullscreen mode Exit fullscreen mode

The lint config is the strictest version (no-emoji). I keep it that way because the goal is acceptance into other awesome-list registries down the line, and they reject any list that fails their own awesome-lint pass.

Keeping it synced with the source

The dataset behind sweepbase.net is a CSV of 141 rows. Five of those are pre-launch products (waitlist, "in development" custody, "TBA" network) — the README rule is "shipping only," so the README count is 136.

The diff between CSV and README runs as a small Node script:

const csvNames = new Set(cards.map(c => c['Card Service'].trim()));
const readmeNames = new Set();
const re = /- \[([^\]]+)\]\(https:\/\/sweepbase\.net\/cards\//g;
let m;
while ((m = re.exec(readme)) !== null) readmeNames.add(m[1].trim());

const inCsvNotReadme = [...csvNames].filter(n => !readmeNames.has(n));
Enter fullscreen mode Exit fullscreen mode

Each time I add a card to the dataset, this tells me what's missing in the README, and I add it manually. Manual is fine because it's once a week at most.

What I underestimated

  1. Alphabetical filter sections. Each region/custody/use-case section repeats card names. Adding one new card means editing 4-5 lists. I have a script in mind but haven't built it.
  2. The "Related Lists" section. The other awesome-lists in the crypto/defi space are mostly stale (2-3 years since update). Including them feels honest but reduces the list's perceived freshness.
  3. Star farming. Two-week organic plan, 23 days later, 1 star. Reality check: the list needs distribution, not just existence.

If you're building an awesome-list, the lint+CI part is fast. The interesting work is keeping it honest as the underlying space changes.

Repo: https://github.com/mbtrilla/awesome-crypto-cards

Top comments (0)