What is a skill card for AI?

A folder with a SKILL.md at its root — an Agent Skill on the open standard — plus the studio file, the card, the tin and the licence. You save it into your own Claude, or load it in any agent that reads the standard, and the AI works one job the studio's way.

Do Lund cards work in ChatGPT, Codex, Cursor or Copilot?

A card is an Agent Skill, and the open standard is read by ChatGPT and Codex, GitHub Copilot and VS Code, Cursor, Gemini CLI, Goose and others; the same folder loads unmodified. The studio has witnessed its cards running in Claude; other hosts earn the mark when a run is witnessed.

How do I save a card as a skill in Claude?

Drop the .skill file into a Claude chat and tap Save skill, or add it under Settings › Capabilities › Skills. It appears in your Personal skills, on by default, and never expires.

Is it a subscription?

No. The Card is one purchase and the files are yours. The Year and the Deck are yearly terms that buy editions; when a term ends nothing you hold stops working — freeze, not loss.

How are the cards scored?

On the published Skill Scale — five criteria, one to five each, to the Hundred: trigger, machinery, law, portability, proof. The reasoning prints under every seal; a WITNESSED mark is earned only by an observed run.

Does anything phone home?

No. The studios are single files that run in your browser offline; the cards are Markdown; every package ships a manifest of hashes so nothing hides inside it.

SKILL CARDS FOR AI · CLAUDE SKILLS · AGENT SKILLS FOR CHATGPT, CODEX, CURSOR AND COPILOT

Skills for Claude — and every agent on the open standard.

What a skill card is — and why you can own one.

A Claude skill is a folder: SKILL.md at the root, the files beside it. Claude reads it and works the way the folder says. Anthropic published that format as the open Agent Skills standard in December 2025, and it is now read by some forty agents — ChatGPT and Codex, Copilot and VS Code, Cursor, Gemini CLI, Goose and more. A Lund card is that folder, cut to a published anatomy, with the studio inside it.

CLAUDE SKILLS

Save a card once and it stays in your Personal skills — on by default, in projects, in Claude Code. Nothing expires.

CHATGPT & CODEX SKILLS

The same folder, unmodified. OpenAI added the standard to ChatGPT and Codex CLI within days of its release.

CURSOR · COPILOT · GEMINI CLI

Any agent on the standard reads SKILL.md the same way; the studio files inside are plain browser files and the scripts plain Python.

WHAT MAKES OURS DIFFERENT

A published anatomy, a manifest of hashes, a stated reach, a licence in plain words, and a score out of 100 with the reasoning printed — earned by a witnessed run, never typed.

Start with a free one.

The Card Forge forges a keeper of your own at no charge. The Funance Free Ledger is a whole set of books, free forever. The seats and the deck are launching soon.

FORGE A FREE CARD→ EVERY CARD, RANKED→

Questions people search.

Are Claude skills the same as Agent Skills?

Yes. Claude skills are Anthropic’s native form of the open Agent Skills standard; a skill written to the spec runs across compatible platforms without modification.

Can I buy a Claude skill and keep it?

Yes. A card is a file; once saved it is yours. The studio never connects to your machine.

Where do I put a skill for Claude Code?

In your project’s skills folder, or install it as a plugin from a marketplace repo. Every card ships ready for either.

Are skills safe to install?

Only ones you can read. An audit found prompt injection in a third of tested skills. Every Lund card is plain Markdown with a manifest of hashes, a stated reach, and a fenced portfolio.

← THE DECK