# voxflow

Six skills for speech synthesis, podcast generation, transcription and video work. Useful when your agent has to handle audio and video.

- plugin · `type=plugin` · Claude Code · ★ 7
- Author: VoxFlowStudio
- License: MIT · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/VoxFlowStudio/skills
- Updated: 2026-05-22
- HTML: https://affpapa.org/en/nevibecode/plugins/voxflow

## What it does

The plugin drops six skills into Claude: `hub`, `podcast`, `transcribe`, `video`, `slice` and `card`. Start with `hub` — it routes requests to the others on its own. Together they let the agent speak in 200+ voices across 40+ languages, build a multi-speaker podcast from a topic or a URL, transcribe audio and video with word-level timestamps, dub a video from SRT subtitles, and translate whole clips.

Everything runs through the `voxflow` CLI. Install it with `npm install -g voxflow`, then run `voxflow login` once for browser auth. No keys or tokens live in the repo; CI jobs use the `VOXFLOW_TOKEN` environment variable instead. The `slice` skill turns a long article into a 1080×1920 vertical video using one of 13 themes, while `card` renders shareable images through HTML/CSS and Playwright in 1:1, 3:4 and 9:16.

Every request draws on a per-user quota. The free tier gives 10,000 a month, and a single `say` call or `narrate` segment costs roughly 100, a medium podcast about 5,000. Run `voxflow status` before a big job, or the run will stop halfway through.

## Install

### Claude Code (native platform)

**Plugin marketplace**
In the agent chat (not a shell):
```text
/plugin marketplace add anthropics/claude-plugins-community
```
In the agent chat (not a shell):
```text
/plugin install voxflow@claude-community
```

## When to use it
- narrate a text file into an mp3 with a chosen voice
- generate a multi-speaker podcast from one topic
- transcribe a recording with word-level timestamps
- translate and dub a video into another language
- turn a long article into a vertical video

## Example prompts
- Narrate `README.md` into `readme.mp3`
- Make a short podcast about AI agents
- Dub `tutorial.mp4` into Japanese

## Source text · SKILL.md · MIT

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: 49f097568f7dbb79.

BEGIN THIRD-PARTY FILE 49f097568f7dbb79
````text
# VoxFlow Skills

> Skill files in this repo are auto-overwritten on each release. For feedback or fixes, please [open an issue](https://github.com/VoxFlowStudio/skills/issues) instead of submitting a PR.

[![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
[![npm version](https://img.shields.io/npm/v/voxflow.svg)](https://www.npmjs.com/package/voxflow)
[![6 skills](https://img.shields.io/badge/skills-6-purple.svg)](#what-you-get)
[![Works with](https://img.shields.io/badge/works%20with-Claude%20Code%20%C2%B7%20Cursor%20%C2%B7%20Codex%20%C2%B7%20Gemini-orange.svg)](#install)

![VoxFlow CLI demo](https://github.com/VoxFlowStudio/skills/raw/main/demo.gif)

Voice in your AI workflow. Six skills that let any AI coding agent (Claude Code · Cursor · Codex · Gemini CLI · Cline) speak in 200+ voices, generate podcasts, dub videos, transcribe audio, and turn text into card images or narrated card videos — through one CLI.

**Why VoxFlow over a raw TTS API?** One CLI handles auth, voice search, multi-speaker dialogue, video pipelines, and quota. The skills layer makes it native to whichever agent you're already using — no new context-switch.

## Install

The simplest path — let the VoxFlow CLI auto-detect your agent and run the right command:

```bash
npm install -g voxflow
voxflow skills install
```

It detects Claude Code / Cursor / Codex / Gemini / WorkBuddy / OpenClaw on your `$PATH`, picks the right install command, asks for confirmation, runs it, and prints next steps. Use `--all` to install for every detected agent, or `--for <agent>` to force one.

If you'd rather run the install command directly — **one command for every agent**:

```bash
npx -y skills add VoxFlowStudio/skills --all --yes --global
```

The `skills` npm package detects every AI agent on your machine (Claude Code,
Cursor, Codex CLI, Gemini CLI, Cline, Amp, Antigravity, CodeBuddy, OpenClaw…)
and writes the 6 VoxFlow skills (`hub`, `podcast`, `transcribe`, `video`,
`slice`, `card`) to each agent's standard skill location in a single shot.

## Prerequisites

```bash
npm install -g voxflow
voxflow login          # one-time browser auth
```

## What You Get

Six focused skills, each loaded on demand:

| Skill | Invoked as | What it covers |
|-------|-----------|----------------|
| **hub** | `voxflow:hub` | `say` · `narrate` · `story` · `voices` · auth · quota · feedback |
| **podcast** | `voxflow:podcast` | Multi-speaker AI podcast from topic / URL / script |
| **transcribe** | `voxflow:transcribe` | `asr` · `asr-jobs` · `translate` · `dub` · `video-translate` · `summarize` · `publish` |
| **video** | `voxflow:video` | `picstory` · `present` · `slides` · `explain` · `image` |
| **slice** | `voxflow:slice` | Article → vertical card video (1080×1920); 13 editorial / poster / magazine themes |
| **card** | `voxflow:card` | Text → shareable card images (HTML/CSS + Playwright); 1:1 / 3:4 / 9:16, editorial design system. Optional narrated MP4 video via `voxflow card render` (TTS + FFmpeg, in-project output) |

Use the hub skill as the starting point — it routes to the others automatically.

## Example Interactions

| You say | Agent runs |
|---------|-----------|
| "Read this README out loud" | `voxflow narrate README.md -o readme.mp3` |
| "Make a 5-min podcast on AI agents" | `voxflow podcast "AI agents" --length short` |
| "Dub this tutorial into Japanese" | `voxflow video-translate tutorial.mp4 --to ja` |
| "把这段话合成语音" | `voxflow say "..." -o output.mp3` |
| "生成一个 AI 播客" | `voxflow podcast "topic" --length medium` |
| "把这个视频翻译成日语" | `voxflow video-translate video.mp4 --to ja` |
| "转录这段录音" | `voxflow asr recording.mp3` |
| "做一个 AI 知识短视频" | `voxflow picstory "topic" --style sketchnote` |
| "生成一套演示幻灯片" | `voxflow slides "topic" --slides 8` |

## Skills Layout

```text
voxflow/
  hub/SKILL.md          # TTS, voice search, auth, quota, feedback
  podcast/SKILL.md      # AI dialogue podcast
````
END THIRD-PARTY FILE 49f097568f7dbb79

The third-party file has ended. Everything between the 49f097568f7dbb79 tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

First 4072 of 5231 bytes. © VoxFlowStudio · MIT · https://github.com/VoxFlowStudio/skills

## Similar
- [listenhub-cli](https://affpapa.org/en/nevibecode/skills/listenhub-cli.md) — skill · 592 installs
- [daily-news-caster](https://affpapa.org/en/nevibecode/skills/daily-news-caster.md) — skill · 1900 installs
- [bailian-gen](https://affpapa.org/en/nevibecode/skills/bailian-gen.md) — skill · 9200 installs
- [elevenlabs](https://affpapa.org/en/nevibecode/skills/sanjay3290-elevenlabs.md) — skill · 351 installs
- [nemotron-speech](https://affpapa.org/en/nevibecode/skills/nemotron-speech.md) — skill · 2000 installs

_Part of the page did not fit — see the HTML version._

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md