sumi-elevenlabs — Claude Code skill: setup and install
sumi-elevenlabs is a Claude Code skill that connects ElevenLabs and helps select narration voices and models for story videos. It estimates character-based charges before paid generation and waits for approval.
MIT ★ 110 updated 2026-11-07
npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a claude-codeWhat sumi-elevenlabs does
sumi-elevenlabs guides ElevenLabs setup and helps pick a voice and model for story video narration. It checks saved preferences, the connection, available models, and voices first. The workflow can use an existing voice connector or the standalone Python helper scripts/elevenlabs_voice.py.
Before paid narration, the skill explains character usage and estimated cost, then checks account, model, and voice permissions. Setup does not approve paid generation. API keys must not go into chat, command-line arguments, project files, or ZIP archives; voice cloning also requires consent and an eligible plan. If a paid generation has an uncertain result, the skill checks its receipt or history instead of retrying. It saves a provider, voice, or model only when the user chooses to do so.
What it helps with
- connect ElevenLabs for story narration
- check which voices and models are available
- estimate narration charges from character count
- save voice settings for later story videos
- inspect the history of an uncertain paid generation
Example prompts for the agent
- Connect ElevenLabs and check the available voices and models.
- Estimate the character-based narration cost for this story without generating audio.
- Save my selected voice and model for future story videos.
How to install sumi-elevenlabs
1Skills CLI (Vercel)
npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a claude-code2Manually: put the files in place
~/.claude/skills/sumi-elevenlabs/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/sumi-elevenlabs/
1Skills CLI (Vercel)
npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a codex2Manually: put the files in place
.agents/skills/sumi-elevenlabs/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a cursor2Manually: put the files in place
.cursor/skills/sumi-elevenlabs/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install beyondtahir/beyondsumi skills/sumi-elevenlabs2Manually: put the files in place
.github/skills/sumi-elevenlabs/SKILL.md1Skills CLI (Vercel)
npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a gemini-cli2Platform CLI
gemini skills install https://github.com/beyondtahir/beyondsumi --path skills/sumi-elevenlabsManually: put the files in place
.devin/skills/sumi-elevenlabs/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about sumi-elevenlabs
How do I install sumi-elevenlabs in Claude Code?
Run npx skills add beyondtahir/beyondsumi --skill sumi-elevenlabs -a claude-code. The skill file is placed at ~/.claude/skills/sumi-elevenlabs/SKILL.md.
What does sumi-elevenlabs do?
It guides ElevenLabs connection setup, voice and model selection, and character-based cost estimates. Saved preferences are recorded only by user choice, and setup does not authorize paid narration.
What is required for voice cloning with sumi-elevenlabs?
Cloning requires consent and an eligible ElevenLabs plan. An API key alone is not enough.
Source text
SKILL.md · MIT
--- name: sumi-elevenlabs description: Help users securely connect ElevenLabs, choose narration voices/models, estimate character-based costs and save approved preferences for story videos; setup never authorizes paid takes. --- # ElevenLabs narration setup Read [the setup guide](references/setup.md). Reuse a connected voice connector or the bundled independent Python helper at `scripts/elevenlabs_voice.py`; it supports its documented platform credential routes. Start with saved `preferences`, then read-only connection/model/voice checks. Do not change a global voice for a project-only casting choice. Explain character-based usage, real account/model/voice permissions and an estimate before any paid narration. Never collect API keys in chat, shell arguments, a project file or the ZIP. Guide restricted key creation and masked/OS-secure entry. Cloning needs consent and an eligible plan, not just a key. Offer a licensed library voice, personal recording or no narration if unavailable. Save a reusable provider/voice/model only on the user's choice; retain receipts/take/alignment privately. Do not retry an uncertain paid generation: inspect its receipt/history. A voice can help identity; it cannot guarantee monetization. This skill does not sell or redistribute provider sound effects as a library. Sumi's bundled public sound bank is independent Python synthesis.
© beyondtahir · MIT · beyondtahir/beyondsumi