cd /news/ai-agents/k-skill-100-skills-that-turn-any-ai-… · home topics ai-agents article
[ARTICLE · art-83845] src=dibi8.com ↗ pub= topic=ai-agents verified=true sentiment=↑ positive

k-skill: 100+ Skills That Turn Any AI Agent Into a Korea Local

NomaDamas released k-skill, an MIT-licensed collection of 100+ agent skills for Claude Code, Codex, and OpenCode that enable AI agents to interact with Korean-specific services such as SRT/KTX train booking, real-time Seoul subway and bike-share data, government filings, and sports scores. The project, which has over 6,800 GitHub stars and was updated on August 1, 2026, requires no separate API layer and can be installed individually or all at once via npx or a Claude Code plugin marketplace.

read6 min views1 publishedAug 2, 2026
k-skill: 100+ Skills That Turn Any AI Agent Into a Korea Local
Image: Dibi8 (auto-discovered)

k-skill is an MIT-licensed collection of 100+ agent skills for Claude Code, Codex, and OpenCode covering Korean-specific services — train booking, real-time subway/bike-share data, government filings, sports scores, and dozens of Korean shopping and life-admin sites — installable individually or all at once.

  • ⭐ 6854
  • TypeScript
  • Python
  • MIT
  • Updated 2026-08-03

42 Real-World OpenClaw Use Casesego lite: The Browser You and Your AI Agents Share

Project thumbnail — from github.com/NomaDamas/k-skill

What Is k-skill? # #

k-skill is a single, MIT-licensed answer to a problem generic AI agents share everywhere: they’re bad at the country-specific, service-specific admin tasks that make up a lot of real daily life. k-skill is a collection of 100+ individually installable skills, each one teaching an agent to interact with one specific Korean service — SRT/KTX train booking, real-time Seoul subway and bike-share data, government legal and business-registration search, sports scores (KBO baseball, KBL basketball, K-League, even LCK esports), and dozens of Korean shopping, delivery-tracking, and life-admin sites.

🔗 GitHub: https://github.com/NomaDamas/k-skill

At 6,800+ GitHub stars, MIT licensed, with a commit from August 1, 2026, and supporting Claude Code, Codex, OpenCode, and OpenClaw/ClawHub, it’s become a reference example of what a deep, single-country skill pack for AI agents can look like — well beyond the usual “weather API” toy example.

No Separate API Layer Required # #

The README is specific about the architecture: no additional client API layer is needed. Where a skill needs a shared, rate-limited credential rather than the end user’s own login, it talks to a lightweight proxy (k-skill-proxy

) over plain HTTP — the agent just calls the skill, and the skill handles routing to the right public API, scraped surface, or proxied endpoint.

Installation # #

Full set #

npx --yes skills add NomaDamas/k-skill --all -g

One skill at a time #

npx --yes skills add NomaDamas/k-skill --skill srt-booking -g

Requires Node.js 18+ and npx

— nothing else.

Claude Code plugin marketplace #

/plugin marketplace add NomaDamas/k-skill
/plugin install k-skill@k-skill

Installed this way, skills are called under a /k-skill:<name>

namespace — for example, /k-skill:lotto-results

.

A Sample of the 100+ Skills, by Category # #

The full list is genuinely long — the README’s own table runs past 100 rows. A representative slice, grouped by what they’re actually for:

Category Example skills
Transportation srt-booking , ktx-booking , express-bus-booking , seoul-subway-arrival , seoul-bike (Ddareungi bike-share), korean-transit-route , highway-traffic-status
Government & legal korean-law-search , nts-business-registration , corporate-registration-consulting , court-payment-order-assistant , korean-patent-search , g2b-order-plan-search (public procurement)
Real estate & finance real-estate-search , housing-official-price , korean-stock-search , k-dart (financial disclosures, 14 endpoints), toss-securities
Civic / real-time city data seoul-density (real-time crowding at 121 hotspots), fine-dust-location , han-river-water-level , emergency-room-beds , ev-charger-nearby
Sports & entertainment kbo-results , kbl-results , kleague-results , lck-analytics (League of Legends esports), korean-cinema-search , ticket-availability (concerts)
Shopping & delivery coupang-product-search , naver-shopping-search , danawa-price-search , delivery-tracking , daangn-used-goods-search (Korea’s Craigslist-equivalent)
Language & documents korean-spell-check , korean-humanizer (flags AI-sounding Korean writing by severity), hwp / rhwp-edit (.hwp document format, Korea’s dominant office format)

Several skills are explicit “thin wrappers” around separately maintained open-source projects rather than reimplementations — korean-jangbu-for

wraps kimlawtech/korean-jangbu-for

for bookkeeping automation, and korean-privacy-terms

wraps the Apache-2.0 kimlawtech/korean-privacy-terms

for generating legally-compliant privacy policies for Korean web apps.

Login Requirements Are Marked Per Skill # #

Every row in the skill table is tagged with whether it needs the user’s own login/credentials. Skills touching real transactions or personal accounts (train/bus booking, job-portal talent search, court filing, brokerage lookups, a restaurant-reservation “sniper” skill) require it. Most public-data lookups — subway arrivals, weather, sports scores, legal statute search, government transaction-price data — need no user credentials at all; some route through the shared k-skill-proxy

for rate-limited public APIs instead.

This matters for anyone evaluating which skills are safe to hand to an agent with less supervision versus which ones should stay behind explicit human approval.

Getting Started, Per the README’s Own Sequence # #

  • Install the full skill set following the install guide - Run the k-skill-setup

skill to resolve credentials and check runtime environment variables - If secrets are missing, follow the security/secrets policyfor the credential resolution order - Install any missing Node/Python packages globally first

  • Open each skill’s own doc for its inputs, examples, and limitations before relying on it

There’s also a ** k-skill-cleaner** meta-skill that combines interview answers with per-agent trigger-count statistics to recommend which installed skills you’re not actually using — a genuinely unusual bit of self-housekeeping for a skill pack this size.

k-skill vs. a Generic Agent Without It # #

Aspect k-skill Generic agent, no skill pack
Korean train/bus booking Dedicated skill per operator Agent has to improvise scraping/API calls each time
Real-time Seoul city data (subway, crowding, bike-share) Purpose-built skills Not available without custom integration
Korean legal/government search Dedicated skills with documented scope Agent guesses at endpoints, likely to hallucinate
Credential handling Documented per-skill (proxy vs. user login) Ad hoc, inconsistent
Coverage breadth 100+ Korea-specific skills Zero out of the box
License MIT N/A

Use Cases # #

1. A Personal Assistant That Actually Knows Korean Services #

Booking an SRT ticket, checking a lotto result, or finding the nearest cheap gas station are all skills an agent can call directly, instead of the agent trying (and often failing) to improvise a scraping approach per request.

2. Building Korea-Market Products on Top of Documented Skills #

The korean-privacy-terms

and korean-jangbu-for

wrappers show a pattern: use k-skill’s skill layer as a documented interface to Korea-specific compliance and bookkeeping logic, rather than re-deriving Korean privacy-law requirements from scratch.

3. Real-Time Civic Data for Local Apps or Agents #

Skills like seoul-density

, han-river-water-level

, and emergency-room-beds

expose real-time Seoul infrastructure data that would otherwise mean separately discovering and integrating each government or municipal API.

4. Auditing Which Skills You Actually Use #

For anyone who installed --all

and ended up with 100+ skills loaded, k-skill-cleaner

is a rare example of a skill pack shipping its own bloat-reduction tool.

Repository Purpose

korean-jangbu-for

wraps as a thin skill layerkimlawtech/korean-privacy-termskorean-privacy-terms

wraps## Related Articles #

42 Real-World OpenClaw Use Cases— for the broader pattern of task-specific agent skills across other domainsego lite: The Browser You and Your AI Agents Share— another skill-based agent extension, for browser automation rather than country-specific service integration

Conclusion # #

k-skill is what a country-specific agent skill pack looks like when someone takes it seriously rather than shipping three demo skills and calling it done: 100+ individually documented, individually installable skills, each with its own login-requirement disclosure and limitations doc, spanning transportation, government, real estate, sports, shopping, and document formats specific to Korea. It’s a strong template for what “make an agent genuinely useful for a specific country’s daily life” can look like beyond generic weather/currency-conversion examples — and directly useful if you’re building for, or living in, Korea.

Best for: Developers building AI agents for Korean users, or Korean users who want their coding agent to handle real Korean life-admin tasks — not a general-purpose international skill pack.

GitHub: https://github.com/NomaDamas/k-skill

Last updated: 2026-08-03

── more in #ai-agents 4 stories · sorted by recency
── more on @nomadamas 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/k-skill-100-skills-t…] indexed:0 read:6min 2026-08-02 ·