Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

21st Ai

ASecurity

Generate and iterate on UI with 21st AI from the terminal via the `21st` CLI (`@21st-dev/cli`): sketch a UI from a prompt, preview the variants in the browser, edit any variant in place with a natural-language change, and pull the final code into the project. Use whenever the user wants to BUILD / SKETCH new UI ("sketch a pricing page", "generate a hero with 21st", "make me a <X> section"), ITERATE on a generated draft ("edit the third variant", "add a toggle to that take"), or GRAB the code ...

207 stars
0 votes
0 copies
0 views
Added 9/4/2026
ai-agentsbashapi

Works with

claude codeclaude desktopcursorterminalcliapimcp

Security Analysis

A100/100

Scanned 9/4/2026

Install to Claude Code

$npx -y skills add NeverSight/skills_feed --skill 21st-ai --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of 21st Ai?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for 21st Ai
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/neversight-21st-ai/badge)](https://www.skillsdirectory.com/skills/neversight-21st-ai)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: 21st-ai
description: >-
  Generate and iterate on UI with 21st AI from the terminal via the `21st` CLI
  (`@21st-dev/cli`): sketch a UI from a prompt, preview the variants in the
  browser, edit any variant in place with a natural-language change, and pull
  the final code into the project. Use whenever the user wants to BUILD / SKETCH
  new UI ("sketch a pricing page", "generate a hero with 21st", "make me a <X>
  section"), ITERATE on a generated draft ("edit the third variant", "add a
  toggle to that take"), or GRAB the code of a draft ("pull that variant's
  code", "сгенери UI на 21st", "поправь третий вариант", "забери код того
  наброска"). For finding/installing existing components use the `21st-cli-use`
  skill; to publish your own, `21st-registry`.
---

# 21st AI — sketch, iterate & grab code from the terminal

21st AI is a fast UI-drafting board. Mental model: **draft it here, then finish
it** — sketch a few variants from a prompt, preview them, refine the one you like
with plain-English edits, and pull its code into the repo. The whole loop runs
from the `21st` CLI (or the matching MCP tools), so you never leave the editor.

Sketch mode produces self-contained **HTML + Tailwind** drafts (fast, no build).
The recommended handoff is the **copy-prompt**: a spec that tells you to rebuild
the design in the target project's real stack — not paste raw HTML verbatim.

## Auth

Same credential as the rest of the CLI: `21st login` (saved token) or a
`21st_sk_` key via `--api-key` / `TWENTYFIRST_TOKEN`. `21st usage` shows tier +
remaining quota. (See the `21st-cli-use` skill for the full auth notes.)

## The loop

```bash
# 1) Sketch — generate variants ("takes") from a prompt. Opens the preview in
#    the browser and prints the projectId.
21st generate "a pricing section with 3 tiers and a monthly/annual toggle"

# 2) List the takes (1-based numbers + whether each has rendered yet).
21st generation <projectId>

# 3) Edit ONE take in place with a natural-language change (PAID — see Metering).
21st iterate <projectId> "make it enterprise-flavored, add an annual toggle" --take 3

# 4) Grab the code to bring into the project (FREE).
21st take <projectId> --take 3            # paste-ready copy-prompt (recommended)
21st take <projectId> --take 3 --code     # raw self-contained HTML instead
```

Open the preview URL (printed by `generate`, and returned by `iterate`) in the
editor's built-in browser to watch the variants render and update in place.

## When to use which

- **generate** — start fresh from a prompt. Returns a preview URL + projectId.
- **generation** — list the takes so you know which `--take N` to act on.
- **iterate** — refine an EXISTING take (edited in place; a new version is
  appended, so you can step back in the workspace). Use this to change something
  already generated instead of starting over.
- **take** — pull a take's code. Default prints the copy-prompt (adapt-to-my-
  stack spec); `--code` prints the raw HTML. Both FREE and unlimited.

## Metering

- **generate** — small free daily bucket, then a paywall (`21st usage` shows
  what's left). Community access / membership lifts it.
- **iterate** — billed by real token cost per edit (the shared 21st AI credit
  pool); it does real model work each call, so batch changes into one clear
  instruction rather than many tiny ones.
- **generation** (listing) and **take** (grabbing code) are **FREE**.

## MCP parity

The same loop is exposed as MCP tools for MCP hosts (Cursor, Claude Desktop,
Claude Code): `generate` (with `mode: "sketch"`), `get_generation`,
`iterate_generation` and `get_take`. `iterate_generation` also returns the
freshly-edited `html` + `copyPrompt`, so you can grab the code without a second
call. Wire up the server with `21st init --client <name> --write`.

## Tips

- A sketch is a **draft, not production code** — treat the copy-prompt as a
  design spec and rebuild it in the project's real components; don't paste the
  HTML as-is.
- Refer to takes by the 1-based number from `generation` / the "Take N" label in
  the preview. `--take` must be a whole number ≥ 1 (a bad value errors rather
  than silently editing take 1).
- Only sketch generations you created can be listed, edited, or fetched.

Attribution

NeverSightNeverSight
View sourceMore from NeverSight →
SSkills DirectorySkills Directory

Your tool, in front of Claude Code builders.

3 founder slots · $299/mo · GSC-verified traffic · sponsors can never buy grades.

See placements

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Your tool, in front of Claude Code builders.

3 founder slots · $299/mo · GSC-verified traffic · sponsors can never buy grades.

See placements

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →