Skip to main content
You use Claude Code or Cursor. This tutorial installs one skill from the public registry — a skill with a measured, verified benchmark, not just stars — and shows it firing in your next session. No account, no SDK, no API key, under 5 minutes. A skill is a SKILL.md file: instructions your agent reads when a matching task comes up (a triage policy, an extraction schema, a commit-message convention). It’s an open format your runtime already understands — DecimalAI’s registry adds the part no file format can: evidence that the skill actually works.
1

Read the scorecard (1 minute)

Open the skill’s public page — no login:app.decimal.ai/skills/support-ticket-kitBefore installing anything, read the two surfaces that make this registry different:
  • Verified lift: +40 pts pass rate — with the skill, the benchmark agent passed 100% of cases; without it, 60%. That’s a with-vs-without A/B run (an ablation) over 25 code-graded cases, executed by the hosted runner — not self-reported by the author. never_hurt means no case got worse with the skill loaded.
  • The fine print that keeps the number honest — the model it was measured on (gemini-3.5-flash), the benchmark date, and the grading method are all on the page. Lift is model-relative: a stronger base model may need a given skill less. The registry shows you the evidence instead of asking you to trust a score.
Also check the safety band: this skill is Passed — a static scan (secrets, remote-code execution, hidden unicode), an AI intent review, and a content review all came back clean. How skills are vetted covers what that does and doesn’t guarantee.
Why this skill for the demo: support-ticket-kit carries a complete internal triage policy for a fictional company. No model can know a company’s internal policy — so when your agent answers correctly, you know the skill fired. It’s a worked example of the general pattern: skills earn lift by supplying knowledge the model can’t have.
2

Install it (pick one of three)

All three routes are anonymous and end in the same place: a SKILL.md on disk where your runtime auto-discovers it.
One command, zero install — Node 18+ is the only requirement:
Launch wave — the npm package is rolling out with the current launch. If npm can’t find decimalai yet, use the CLI-pull or web tab and check back shortly.
This fetches the skill from the public registry and writes .claude/skills/support-ticket-kit/SKILL.md (plus any bundled attachments), stamped with provenance frontmatter (source + source_sha256) and pinned in .decimal/skills.lock.json so you can always tell exactly which version you’re running.Using Cursor or another .agents/skills/-scanning runtime? Point it there:
3

Watch it fire

Start a new Claude Code (or Cursor) session in the project — skills are discovered at session start. Then paste a task the skill was built for:
The skill’s policy says VIP-client tickets that request access get AUTO_REJECT — they must go through the client’s external vendor portal, not internal IT. That’s deliberately counterintuitive: without the skill, every model reads “VIP + urgently” and answers some flavor of urgent.
What you should see:
  • The transcript shows the agent reading support-ticket-kit (Claude Code surfaces the skill it loaded, or you’ll see it open .claude/skills/support-ticket-kit/SKILL.md).
  • The answer is AUTO_REJECT, citing the VIP/access rule — not “urgent.”
  • If you get an urgent-style label instead, the skill didn’t load: check the file path and frontmatter, start a fresh session, or name it explicitly — “use the support-ticket-kit skill to triage this ticket.”

​
What just happened

You installed a plain-markdown skill your runtime already knows how to read, and you picked it off a scorecard instead of a download count. The policy in this particular skill is a worked example — fork it and swap in your rules before using it for real work (the benchmark methodology carries over; the fictional policy shouldn’t).

​
Next steps

All the no-SDK routes

Web copy, CLI pull, npx, raw URLs, the MCP server, and the Claude Code plugin — one page.

Build a whole agent from skills

Assemble a billing & order support agent from five measured registry skills — and wire it so usage counts.

Browse the registry

Every published skill, ranked by measured effectiveness. No login to browse.

How the numbers are made

skillevaluation is the open spec + runner behind every verified lift number. Re-run any benchmark yourself.