Most people run Claude like a smarter search box: "do some research", "write me a post". But you can assemble a whole content studio out of it — if you bolt on the right skills 🔥
What it is and who it's for
Not a single tool but a set of 6 Agent Skills that turn Claude Code into a content pipeline: from research and competitor ad teardowns to writing in your voice, voiceover and video. The skills come from mixed sources — some official (ElevenLabs, Remotion), some from the big community repo ComposioHQ/awesome-claude-skills (~67.7k★, Apache-2.0).
Who it's for: people already running content through Claude Code who are tired of hopping between five services. The idea isn't "Claude as search" but "Claude as a studio": one agent holds the whole pipeline and your style.
The 6 studio skills
- Deep Research — autonomous multi-step research on a topic (the ComposioHQ implementation orchestrates the Gemini Deep Research Agent): market, competitors, facts with sources.
- Content Research Writer — writes articles and posts: research, sourced quotes, hook polishing, section-by-section feedback — and keeps your authorial voice.
- Competitive Ads Extractor (called "Competitor Ads Extractor" in the viral post) — pulls competitor ads from the ad library (Facebook/LinkedIn), takes screenshots and breaks them down: which pains, offers and creatives land.
- Voice DNA — analyzes your texts and builds a JSON voice profile: tone, rhythm, sentence length, vocabulary, stop words. After that the agent writes "like you", not generic AI.
- ElevenLabs (official skill) — voiceover for text: text-to-speech with ElevenLabs voices straight from the agent, plus sound effects, music and voice-changer.
- Remotion (the post says "Re-motion"; the real name is Remotion) — video rendered from code in React: the agent writes a composition and assembles a vertical clip from the script.
Net result: research → text in your voice → voiceover → video, without leaving the agent.
How to set it up
The official ones install with a single command:
npx skills add elevenlabs/skills # voiceover, needs ELEVENLABS_API_KEY
npx skills add remotion-dev/skills # video in React
The ComposioHQ skills and Voice DNA are installed by copying: drop the skill folder into ~/.claude/skills/ (in the ComposioHQ/awesome-claude-skills repo those are the deep-research, content-research-writer, competitive-ads-extractor folders; Voice DNA comes from an open implementation repo, e.g. az9713/ai-co-writing-claude-skills).
After ElevenLabs, set the key: export ELEVENLABS_API_KEY="...". Check that the agent sees the skills — then run the pipeline (prompt below).
When to use it, when not to
✅ Take it if you already do content through Claude Code and want one pipeline research → voice → voiceover → video, with no service-hopping. ✅ Take it if you need a reproducible "brand voice": Voice DNA + Content Research Writer hold the style from post to post. ❌ Skip Remotion if you're not willing to touch code → use a visual editor (CapCut, ready-made templates): Remotion is a React code framework, not a point-and-click editor. ❌ You only need research and not a whole studio → use a ready Deep Research on the web (see the "Deep Research" and "Gemini Deep Research" cards) instead of wiring up skills. ❌ You want one skill built for your own pipeline rather than someone else's set → build it with skill-creator (card in this section).
The honest price
- Deep Research, Content Research Writer, Competitive Ads Extractor — from ComposioHQ/awesome-claude-skills, Apache-2.0, free. But Deep Research in this version calls Gemini Deep Research, so your Google/Gemini access is in play.
- Voice DNA — an open community skill, free.
- The ElevenLabs skill is official and free, BUT voiceover is billed in credits from your ElevenLabs account: the free plan gives 10,000 credits/month (~10 minutes of speech on Multilingual v2) and no commercial license; commercial use and volume are paid (plans from $5/month).
- Remotion is official but source-available (not classic OSS): a Free License for solo devs and teams up to 3 people, commercial use allowed; from 4 people on it's a paid Company License (Creators $25/month per seat; Automators $0.01 per render, minimum $100/month). Plus rendering heavy video eats CPU and time. Bottom line: you really can assemble the studio for 0 ₽, but voiceover beyond the ElevenLabs free limit and Remotion on a team of 4+ are paid. The free/oss tag is about the skills themselves, not the services underneath.
Gotchas
- You need an agent that supports Agent Skills (Claude Code and compatible ones). These are skills, not MCP and not plugins.
- Mixed sources: the official ones (ElevenLabs, Remotion) install via
npx skills add, the community/Composio ones by copying a folder. There's no single "install everything" button. - Reels routinely mangle the names: "Re-motion" = Remotion (remotion.dev), and "Competitor Ads Extractor" is competitive-ads-extractor in the repo. Install by the exact name.
- Deep Research leans on an external Gemini Deep Research Agent — your own keys and limits, not something "built into Claude".
- Voice DNA isn't a single official product but a pattern with several implementations; check whose repo you're installing and what the skill does with your texts.
- Competitive Ads Extractor scrapes ad libraries — follow their rules and privacy expectations, don't break ToS.
- The ElevenLabs free plan has no commercial license: that voiceover can't go into monetized content.
- The agent genuinely writes files and burns tokens/credits — keep git around and review what it generates.
🤖 Prompt accelerator
"You are my content studio on Claude Code with 6 skills. Topic:
, platform: <IG/YouTube/blog>. 1) Deep Research + Content Research Writer: gather research with sources and pull out 3 hooks that work. 2) Competitive Ads Extractor: break down ads from in the ad library — which pains and offers land. 3) Voice DNA: write strictly in my voice from the profile <voice-dna.json>. 4) Generate a 40-second script for the platform. 5) ElevenLabs: voice it with . 6) Remotion: assemble a vertical 1080×1920 video from the script. After each step show me the result and wait for "ok" before moving on."