fralapo/awesome-agent-skills

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 — including OpenAI Codex CLI , which this repo also ships ready-to-discover folders for.
## Contents
What's in here
Install
Plugin marketplace (recommended)
Manual (symlink or copy)
Codex CLI (OpenAI)
How a skill is organized
How I build them
Contributing
License
## What's in here
Eighteen 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 2.0/2.5 video gen: storyboards, camera-move vocab, multi-modal refs, identity locks, POV/bodycam, found-footage, stop-motion, game-UI HUD. | ~1000+ prompts/cases across 4 awesome-lists | seedance, bytedance video, text-to-video prompt |
| image-gen-prompts | Prompt engineering for any major image model (Midjourney, GPT Image 2.5 Flare/Sunburst, FLUX, SDXL, Seedream, Imagen, Ideogram...). 40+ templates, 20 edit-repair recipes, auto-routes by model named. | ~19,170 prompts across 14 sources | image prompt, nano banana, gpt image 2.5, 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, YouTube energy and sponsor-readiness. | Practitioner field notes + tested operational frameworks | personal brand, instagram algorithm, organic acquisition, hero's journey, manychat, youtube |
| 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 |