Curated collection of AI agent skills by Citedy — works with Claude Code, Codex, Gemini CLI, Droid, and any AI coding agent that reads .md skill files.
This package is actively maintained. We regularly add new skills and improve existing ones. Run
npx @citedy/skills updateto get the latest version at any time.
npx @citedy/skills installThat's it. All skills and slash commands are copied into your project's .claude/ directory, ready to use.
npx @citedy/skills install youtube tiktoknpx @citedy/skills@latest updateNew skills are added regularly. Use @latest to make sure you always get the newest version — no stale cache.
npx @citedy/skills list| Name | Slash Command | Description | Requires |
|---|---|---|---|
youtube |
/any-youtube |
Extract structured metadata from any public YouTube video | AnyCrawl API |
instagram |
/any-instagram |
Extract public Instagram profile, post, and reel data | AnyCrawl API |
tiktok |
/any-tiktok |
Extract public TikTok video, profile, and hashtag data | AnyCrawl API |
social |
/any-social |
Auto-detect platform and route to the correct extractor | AnyCrawl API |
| Name | Slash Command | Description |
|---|---|---|
schema |
/schema-markup |
Add, fix, or optimize schema.org JSON-LD structured data |
icons |
/icon-design |
Select semantically appropriate icons (Lucide/Heroicons/Phosphor) |
domains |
/domain-hunter |
Search domains, compare prices, find promo codes |
skill-eval |
/skill-eval |
Validate slash command quality (frontmatter, jargon, descriptions) |
symphony |
/codex-symphony |
Bootstrap local OpenAI Symphony + Linear orchestration for any repo |
token-usage |
/token-usage |
Analyze Claude Code token consumption and estimated costs |
prompt-analyzer |
/prompt-analyzer |
Analyze prompts for constraint complexity, audit risks, and generate optimized rewrites for Claude/GPT |
| Name | Slash Command | Description |
|---|---|---|
spawning-plan |
/spawning-plan |
Design and spawn optimal agent teams |
code-review |
/code-review-team |
Parallel multi-agent code review with 4 reviewers |
Agent Team skills require Claude Code with Agent Teams support (TeamCreate, TaskCreate, SendMessage).
Extraction skills use the AnyCrawl Scrape API for LLM-powered structured data extraction.
Sign up at anycrawl.dev and get your API key.
# .env.local (or .env)
ANYCRAWL_API_KEY_DEV=ac-your-key-herenpx @citedy/skills installOnce installed, Claude Code automatically picks up skills when you provide a social media URL. Or invoke directly:
/any-youtube https://www.youtube.com/watch?v=VIDEO_ID
/any-instagram https://www.instagram.com/username/
/any-tiktok https://www.tiktok.com/@user/video/VIDEO_ID
/any-social <any-supported-url>
Each skill returns a structured JSON artifact saved to /tmp/ with:
normalized— Platform-specific structured data (title, stats, channel/author, description, etc.)anycrawl— Raw AnyCrawl response with markdown contentrequest— The exact API payload used
- Node.js 18+
- AnyCrawl API key
- Claude Code or Codex
The prompt-analyzer skill applies findings from "How LLMs Follow Instructions: Skillful Coordination, Not a Universal Mechanism" (Rocchetti & Ferrara, 2026) to audit and optimize prompts.
What it does:
- Decomposes prompts into constraint types (structural, lexical, semantic, stylistic)
- Scores complexity: Constraint Density, Cross-Type Mixing Index, Order Risk
- Audits for conflicts, ambiguity, implicit constraints, and model-specific risks
- Generates optimized rewrites — separate variants for Claude and GPT
Usage:
/prompt-analyzer path/to/prompts.md
The file can contain multiple prompts separated by ---. Works with plain text, system prompts, and structured files (JSON/YAML).
Install just this skill:
npx @citedy/skills install prompt-analyzerThe symphony skill is different from the extraction skills. It bootstraps local OpenAI Symphony orchestration into the current repository and installs:
WORKFLOW.symphony.mdscripts/symphony/*.env.symphony.examplecodex-symphonywrapper in~/.local/bin
Install just that skill:
npx @citedy/skills install symphonyThen in your target repo:
/codex-symphonyYou will still need your own:
LINEAR_API_KEYLINEAR_PROJECT_SLUGSOURCE_REPO_URLSYMPHONY_WORKSPACE_ROOT- optional
GH_TOKEN
MIT — Citedy