A small, opinionated set of agent skills I actually use.
Each one is mined from a real corpus (community awesome-lists, official docs, books, field notes), anywhere from a couple hundred examples to tens of thousands. None of it written from imagination.
Built on Claude Code's skill format, installable as a plugin and usable by any agent that reads SKILL.md.
Ten skills. Each is a single folder with a SKILL.md (frontmatter + workflow) plus deep-dive material (a references/ directory, or chapters/ + cheatsheet/patterns/glossary for the book-distilled ones). The agent (Claude Code, or any tool that reads the same format) matches the description field against whatever you're asking and loads the skill only when it's relevant. No eager bloat.
| Skill | What it does | Corpus mined | Typical trigger |
|---|---|---|---|
seedance-prompts |
Prompt templates for ByteDance Seedance 2.0 video gen: storyboards, camera-move vocab, multi-modal refs. | ~195 prompts across 3 awesome-lists | seedance, bytedance video, text-to-video prompt |
image-gen-prompts |
Prompt engineering for any major image model (Midjourney, GPT Image, FLUX, SDXL, Seedream, Imagen, Ideogram...). 28+ templates, auto-routes by model named. | ~18,900 prompts across 10 sources | image prompt, nano banana, midjourney, flux, sdxl, ideogram, dall-e, mockup |
awesome-readme |
Write GitHub READMEs that land on awesome-lists and sound human, not AI-generated. 8 templates + badge/GIF recipes + anti-slop audit pass. | Best-README-Template + ~100 awesome-readme entries + Standard-Readme | readme, awesome readme, github readme, humanize readme |
llm-wiki |
Persistent LLM-maintained knowledge base in an Obsidian vault, Karpathy's "compile, don't retrieve" pattern. ingest/query/lint ops, hallucination spot-checks. |
Karpathy's llm-wiki gist + 3 production Obsidian-wiki repos + field-tested corrections | obsidian, llm wiki, karpathy wiki, second brain, pkm, wiki ingest |
creative-director |
Router into 5 role toolkits (copywriter, idea generator, creative leader, visionary, evaluator), each loading only its own book digest. | 16 books + smixs evaluation engine + 23 TED talks | creative director, headline, copywriting, evaluate concept, typography, visual literacy |
social-algorithm |
Organic personal-brand growth + Instagram algorithm dynamics: indirect selling, acquisition funnels, hook patterns, delegation systems. | Practitioner field notes + tested operational frameworks | personal brand, instagram algorithm, organic acquisition, hero's journey, manychat |
ffmpeg |
Full FFmpeg CLI reference: codecs, formats, filtergraphs, protocols, capture devices, plus full official docs bundled for grep. | Official FFmpeg documentation, all tools & libraries | ffmpeg, ffprobe, transcode, filtergraph, -map, hls, crf, nvenc |
video-editor |
Software-agnostic decision system for editing viral video: hook system, pacing, motion, captions, sound design, transitions. | 25 expert YouTube video-editing breakdowns, deduped | video editing, reel, retention, hook, pacing, b-roll, motion graphics |
public-speaking-persuasion |
Public speaking + persuasion theory: speech structure, logical-fallacy/cognitive-bias catalogs, objection handling, pitching. | 51 theoretical transcripts on communication and persuasion | public speaking, persuasion, pitch, logical fallacy, objection handling |
marketing-mba |
MBA-level marketing + startup execution: segmentation, pricing, value proposition, sales pipeline, negotiation, founder frameworks. | 90 theoretical marketing/startup/business transcripts | marketing, value proposition, pricing strategy, go-to-market, negotiation, porter five forces |
More coming when I build one that earns its place. I'd rather ship seven that work than twenty that almost do.
One command set, every skill, native on Windows (no symlink permissions needed), updates via /plugin. Run inside Claude Code:
/plugin marketplace add fralapo/awesome-agent-skills
/plugin install awesome-agent-skills@awesome-agent-skills
That installs the whole collection as a single plugin, so all skills become available at once. Update later with:
/plugin marketplace update awesome-agent-skills
/plugin update awesome-agent-skills@awesome-agent-skills
Prefer just one skill? Use the manual method below and symlink/copy only that folder.
Pick one. Replace <skill> with the skill name (seedance-prompts, image-gen-prompts, awesome-readme, llm-wiki, social-algorithm, creative-director, ffmpeg, video-editor, public-speaking-persuasion, or marketing-mba).
Commands below target Claude Code's default skills dir (~/.claude/skills/). Other agents: swap the destination for whatever path your tool reads (commonly ~/.config/<agent>/skills/ or .agent/skills/ in a project root).
macOS / Linux:
git clone https://github.com/fralapo/awesome-agent-skills.git ~/src/awesome-agent-skills
ln -s ~/src/awesome-agent-skills/skills/<skill> ~/.claude/skills/<skill>Windows (PowerShell, admin or developer mode):
git clone https://github.com/fralapo/awesome-agent-skills.git $env:USERPROFILE\src\awesome-agent-skills
New-Item -ItemType SymbolicLink `
-Path "$env:USERPROFILE\.claude\skills\<skill>" `
-Target "$env:USERPROFILE\src\awesome-agent-skills\skills\<skill>"macOS / Linux:
git clone https://github.com/fralapo/awesome-agent-skills.git
cp -r awesome-agent-skills/skills/<skill> ~/.claude/skills/Windows:
git clone https://github.com/fralapo/awesome-agent-skills.git
xcopy /E /I awesome-agent-skills\skills\<skill> $env:USERPROFILE\.claude\skills\<skill>git clone https://github.com/fralapo/awesome-agent-skills.git ~/src/awesome-agent-skills
for d in ~/src/awesome-agent-skills/skills/*/; do
ln -s "$d" ~/.claude/skills/
doneRestart the agent after install. Most tools load skills at startup.
skills/<skill-name>/
├── SKILL.md # required: frontmatter (name, description) + workflow
└── references/ # optional: deep-dive files loaded only when relevant
├── patterns.md
├── templates.md
└── examples.md
Some skills carry extra on-demand files alongside or instead of references/: a glossary.md / patterns.md / cheatsheet.md trio (creative-director, ffmpeg), a chapters/ directory (ffmpeg), or roles/ + books/ (creative-director). Same idea: depth loaded only when the agent needs it.
The frontmatter description field is the matcher. Keep it specific and keyword-dense; the agent uses it to decide whether to load the skill at all.
The whole repo also ships as one installable plugin; see Install. The plugin manifests live in .claude-plugin/ (plugin.json bundles every skill, marketplace.json lists it).
The pattern is consistent across every skill in this repo:
- Mine a real corpus. Community awesome-lists, vendor docs, internal notes. Several hundred examples minimum. I don't write skills from imagination.
- Cluster what actually recurs. Templates, anti-patterns, category-specific moves. Things a real prompter or writer reaches for ten times a week.
- Write
SKILL.mdshort. Frontmatter, when-to-use, quick rules, pointers into references. The agent reads this at load time; every extra paragraph costs context. - Push depth into
references/. Topical files. Loaded on demand when the agent follows a link. - Cite sources. When a template or pattern came from a specific public repo, it gets named in
examples.mdor the frontmatteranalyzed_reposlist.
If you're writing your own agent skills, those five steps are the only ones that have mattered for me. Everything else is decoration.
This is my personal collection. Every skill here I author and mine myself. I'm not looking to absorb new skills from outside (vendor-tool wrappers especially aren't a fit). What does help:
- Bug fix: the smallest diff that fixes the thing.
- Improvement to an existing skill: cite the specific pattern or source that motivated it.
Open an issue first. If you've built your own skill, I'd rather link to your repo than fold it into mine.
MIT.