rnskill

August 25, 2026 · View on GitHub

AI Agent Skills by 雪踏乌云 for Codex, Claude Code, and other SKILL.md-compatible agents.

Currently 56 skills in two categories:

🎬 Content Creation (Video / Image / Article)

With this skill set + Codex + HyperFrames + HeyGen + IndexTTS2, I spent less than 10 hours total over the past month — grew 2k followers on Douyin and landed my first paid brand deal. Covers topic selection, scriptwriting, AI voice cloning, digital avatars, video editing, subtitles, cover design, article-to-image, production orchestration, motion graphics, and business diagnostics.

💻 Coding & Development

General-purpose coding skills, more coming soon.

RN Cover Skill

rn-cover-skill creates 5:2 editorial diagram covers from a title and theme—no reference image required. ImageGen invents a new right-side concept for every cover, while the compositor controls exact typography, the warm-white canvas, and editable SVG output. The result keeps a recognizable visual family without repeating one reference layout.

RN Cover Skill example

Exact #FAF9F5 warm-white canvas with Chinese and mixed-language headline priority
Vertically centered left text block; the right visual yields to headline length
A thumbnail-visible fine grid anchors the right field; its start, spacing, strength, and artwork position remain adjustable
Fresh isolated artwork per cover, delivered as self-contained SVG and PNG

Example:

Use $rn-cover-skill to create a 5:2 cover titled "Open-source Claude-style Cover Skill"

Requirements

Codex, Claude Code, or another Agent that supports project-level skills
A target project that can load .agents/skills/<skill-name>/SKILL.md

Installation

Claude Code Plugin Marketplace

claude plugin marketplace add Pluviobyte/rnskill
claude plugin install ra-人话@rnskill

Universal (Codex / Claude Code)

npx -y skills add Pluviobyte/rnskill -g --all

Or install a single skill:

npx -y skills add Pluviobyte/rnskill --skill ra-人话

Manual Install

# Codex
mkdir -p <project>/.agents/skills
cp -R skills/ra-人话 <project>/.agents/skills/ra-人话

# Claude Code
mkdir -p <project>/.claude/skills
cp -R skills/ra-人话 <project>/.claude/skills/ra-人话

All Skills

Skills marked ⬡ are from or adapted from external open-source projects — see the Source column and Credits below.

Agent & Tool Invocation

Skill	Description	Source
`grok-build-cli`	Let Codex invoke the local Grok Build CLI: verify login and models, choose single-turn or agentic mode, monitor long calls, and return results reliably	Original

Topic & Planning

Skill	Description	Source
`ra-选题`	Full topic lifecycle: create cards, deepen research, recommend (benchmark + own data), route to video/article/image workbench	Original
`ra-实操策划`	Practical long-form planning: test prompts, timeline structure, on-camera script, screen-recording checklist	Original
`ra-hook$	\text{Short} \text{video} \text{hook} \text{selection}: 7 \text{types} \times 3 \text{categories}, \text{with} \text{prerequisites}, \text{templates}, \text{and} \text{common} \text{mistakes}	\text{Original}
$ra-video-title`	Video title generation: lock theme → analyze benchmarks → 8-12 two-part candidates + Top 3 picks	Original

Content Writing

Skill	Description	Source
`ra-video-wash-pipeline`	Video script-wash orchestrator: download → transcript → rewrite → QC → queue for production	Original
`ra-逐字稿提取skill`	Extract verbatim transcript from Douyin/Xiaohongshu videos via watermark removal + Paraformer ASR	Original
`ra-洗稿`	Script rewrite: chains ra-人话 → dbs-ai-check → dbs-hook → dbs-resonate → ra-video-title	Original
`ra-人话`	Chinese de-AI writing: bans binary contrast shells, fake insight markers, lecture colons; preserves author judgment	Original
`ra-公众号提取`	WeChat article full-text extraction via MicroMessenger UA spoofing, stdlib only	Original

Video Download

Skill	Description	Source
`ra-video-download`	Download from Douyin/YouTube/Bilibili/Twitter/Xiaohongshu via yt-dlp + TikHub	Original

Voice

Skill	Description	Source
`tts-skill`	Local IndexTTS2 voice cloning with locked lossless reference, no cloud fallback, voice_manifest.json audit trail	Original

Digital Avatar

Skill	Description	Source
`heygen-digital-avatar`	HeyGen Digital Twin generation & compositing: CLI OAuth, circle crop layout, audio approval gate	Original

Video Editing

Skill	Description	Source
`ra-local-talking-head-cut`	Local talking-head rough cut: ASR → pre-cut review → user approval → semantic edit → loudness/cut QC	Original
`video-use`	Conversational video editing: transcribe, cut, color grade, overlay animations, burn subtitles	⬡ Browser Use · MIT
`ai-jian-koubo`	Talking-head transcription + AI stutter detection + waveform review UI + FCPXML export	⬡ Inspired by chengfeng/videocut-skills, rewritten · AGPL-3.0
`chengfeng-videocut-skills`	Chengfeng's original talking-head cutter: transcription → stutter detection → review page → FCPXML	⬡ chengfeng / AI产品自由 · Apache-2.0

Subtitles

Skill	Description	Source
`ra-audio-to-subtitles`	Volcengine Doubao-ASR word-level timestamps + alignment/fragment/connector/reading-speed QC	Original
`skill-captions`	Subtitle styling & burn-in: anchor-dark / anchor-light styles, 4K native rendering, render QC	Original

Visual & Cover

Skill	Description	Source
`ian-xiaohei-illustrations`	Ian's original Xiaohei article illustrations: white hand-drawn 16:9 visuals with restrained red/orange/blue annotations for judgments, processes, and metaphors	⬡ Original author Ian / helloianneo · MIT
`skill-cover`	Cover generation with registered styles and dual-ratio assets	Original
`editorial-dot-cover`	Editorial dot-grid cover: warm gray paper + oversized Chinese title + whitespace + dotted vector icon → SVG + PNG	Original
`editorial-collage-motion`	Halftone paper collage motion: decode reference → generate stills → assemble-from-empty animation	⬡ Inspired by Vikash Kumar / Arcads Collage Motion, local free edition
`rn-niulai-style-image`	Rebuild photos or film stills in the observed rough inflatable-suit 3D grammar of 《牛来》. Default is film-3D; ink-poster mode is explicit only	Original

Article to Image

Skill	Description	Source
`xhs-article-to-images`	Markdown article → Xiaohongshu 3:4 image cards, 5 design skins + 3 feminine themes	Original

Production & QC

Skill	Description	Source
`ra-video-production-director`	Production director: reads handoff contract → dispatches downstream skills → manages state/archive/QC	Original
`ra-复盘`	Content review: collect data → viral grading (R/M) → attribution → asset archiving → topic card writeback	Original

Motion Graphics (HyperFrames)

Skill	Description	Source
`rn-motion-director`	Motion-first AI video director: topic → motion concept, visual metaphors, beat graph, anti-PPT QC	Original
`rn-motion-replica`	Reference motion study: builds original editable HyperFrames project from authorized reference + QC	Original
`rn-human-motion-extractor`	Human-motion extraction: reference video → frame-level body/hand keypoints, anonymized skeleton video, and confidence QC	Original
`rn-dark-saas-video`	Dark cinematic SaaS product video: 8 scene blueprints, 3 timing presets	Original
`rn-bw-text-opener`	Black-white typed text opener with synced SFX, 3 timing presets, Python timing planner	Original
`rn-replica-qc`	Replica QA: 5 fidelity levels + asset/runtime/delivery full-frame gates	Original
`rn-cover-skill`	Reference-free editorial diagram covers with adaptive left text, fresh right-side artwork, editable SVG + PNG	Original

dbs Business Toolkit (22 skills)

From @dontbesilent's open-source dbskill, CC BY-NC 4.0. Used as automated quality gates in the video production pipeline; also available standalone.

Skill	Description
`dbs`	Main router: pre-task routing + post-task navigation
`dbs-hook`	Hook diagnosis: pairs with ra-hook (ra-hook picks type, dbs-hook polishes execution)
`dbs-resonate`	Draft resonance diagnosis via communication psychology framework
`dbs-ai-check`	AI writing fingerprint detection
`dbs-content`	Content creation diagnosis
`dbs-spread`	Transmission psychology decoder: 5 communication theories
`dbs-diagnosis`	Business model diagnosis: consultation + checkup modes
`dbs-benchmark`	Benchmark analysis: five-filter method
`dbs-goal`	Goal clarification via Wittgenstein's philosophy of language
`dbs-deconstruct`	Concept deconstruction to atomic level
`dbs-action`	Execution block diagnosis via Adlerian psychology
`dbs-slowisfast`	Slow-is-fast diagnosis: find seemingly slower methods that build lasting assets
`dbs-good-question`	Fuzzy problems → agent-solvable problem briefs
`dbs-learning`	Interactive learning with feedback-driven depth/pace adjustment
`dbs-chatroom`	Topic-based expert chatroom with multi-role dialogue
`dbs-chatroom-austrian`	Austrian economics chatroom: Hayek × Mises × Claude
`dbs-content-system`	Content structuring system: turn archives into reusable assets
`dbs-decision`	Personal decision system: local knowledge project for long-running domains
`dbs-xhs-title`	Xiaohongshu article titles: 75 proven viral formulas
`dbs-save`	Save diagnosis state to disk
`dbs-restore`	Restore last saved diagnosis state
`dbs-report`	Package multiple diagnoses into a deliverable Markdown report
`dbs-agent-migration`	Agent workspace migration: Claude Code / Codex / Grok three-host consistency