DojoCodingLabs/remotion-superpowers

{   "$schema": "https://anthropic.com/claude-code/marketplace.schema.json",

"$schema": "https://anthropic.com/claude-code/marketplace.schema.json",

},       "source": "./",       "homepage": "https://github.com/DojoCodingLabs/remotion-superpowers",

"source": "./",       "homepage": "https://github.com/DojoCodingLabs/remotion-superpowers",       "category": "media",

"homepage": "https://github.com/DojoCodingLabs/remotion-superpowers",

"author": {     "name": "Dojo Coding Labs",     "url": "https://dojocoding.io"

"name": "Dojo Coding Labs",     "url": "https://dojocoding.io"   },

"url": "https://dojocoding.io"

},   "homepage": "https://github.com/DojoCodingLabs/remotion-superpowers",   "repository": "https://github.com/DojoCodingLabs/remotion-superpowers",

"homepage": "https://github.com/DojoCodingLabs/remotion-superpowers",

"repository": "https://github.com/DojoCodingLabs/remotion-superpowers",

[Credential-related source omitted]

[Credential-related source omitted]

- Install `@remotion/captions` with `npx remotion add @remotion/captions` if needed. - Copy the element for the chosen style into the project from `https://www.remotion.dev/elements/captions/<element>.md`: Popping Word Captions for TikTok, 

- Copy the element for the chosen style into the project from `https://www.remotion.dev/elements/captions/<element>.md`: Popping Word Captions for TikTok, Word Highlight Captions for karaoke, Basic Captions for YouTube and minimal.

[Credential-related source omitted]

[Credential-related source omitted]

- Limits: sync analysis up to 1 hour, async up to 2 hours, direct uploads up to 4 GB - Index only for `search`; an index is reusable across many searches, an asset is per analysis (delete it afterwards) - For anything else, see the official

- Index only for `search`; an index is reusable across many searches, an asset is per analysis (delete it afterwards) - For anything else, see the official docs: https://docs.twelvelabs.io (and `llms.txt`) - Combine analyzed footage with st

- For anything else, see the official docs: https://docs.twelvelabs.io (and `llms.txt`)

[Credential-related source omitted]

Get the renditions with `pexels_get_video` (`id`) and download the closest one into `public/footage/` (`mkdir -p public/footage`, then `curl -L -o`). Download only the clips you'll use. Pexels doesn't require attribution, but it's appreciat

1. Transcribe each voiceover file with the official word-level flow (`remotion-captions/transcribe-captions.md`, `@remotion/whisper-webgpu`). Without a usable GPU, use ElevenLabs Scribe instead (official `speech-to-text` skill, `scribe_v2` 

[Credential-related source omitted]

[Credential-related source omitted]

Get the renditions with `pexels_get_video` (`id`), take the file link closest to the composition size and download it: `mkdir -p public/footage && curl -L -o public/footage/<name>.mp4 <link>`. Download only the clips you'll use. Pexels does

# Download each selected clip curl -L -o public/footage/[descriptive-name].mp4 "[video_file_link]"

curl -L -o public/footage/[descriptive-name].mp4 "[video_file_link]"

mkdir -p public/footage curl -L -o public/footage/clip-[descriptive-name].mp4 "<output URL>"

curl -L -o public/footage/clip-[descriptive-name].mp4 "<output URL>"

mkdir -p public/images curl -L -o public/images/img-[descriptive-name].png "<output URL>"

curl -L -o public/images/img-[descriptive-name].png "<output URL>"

TwelveLabs has no Agent Skills; its agent docs are at https://docs.twelvelabs.io (and `llms.txt`).

[Credential-related source omitted]

[Credential-related source omitted]

- **remotion-media tools write to `public/<output_name>.<ext>`** in the directory the agent was launched from, so the agent must be started from the Remotion project root. They can't write into subfolders, so encode the type in the name: `v

- **Files you download yourself** (Pexels, Replicate outputs) go into subfolders: `mkdir -p public/footage public/images`, then `curl -L -o public/footage/<name>.mp4 <url>`.

| Captions you transcribe (`captions-workflow.md`) | `public/<source>.captions.json`, and `public/<source>.srt` if exported | Named after the source file | | Files you download yourself (Pexels, Replicate) | `public/footage/` and `public/im

| Hosted SFX (`https://remotion.media/*.wav`) | Nowhere; use the URL as `src` | — |

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

| Voiceover | `vo-scene1.mp3`, `vo-scene2.mp3`, … | 1.0 | One `<Audio>` per scene, inside that scene | | Background music | `music-bg.mp3` | 0.1–0.2 under voiceover, 0.25–0.4 without | Composition level, with fade in/out and ducking | | SFX

| Background music | `music-bg.mp3` | 0.1–0.2 under voiceover, 0.25–0.4 without | Composition level, with fade in/out and ducking | | SFX (transition, impact) | `sfx-whoosh.mp3` or a hosted `https://remotion.media/*.wav` | 0.5–0.8 | `from` 

| SFX (transition, impact) | `sfx-whoosh.mp3` or a hosted `https://remotion.media/*.wav` | 0.5–0.8 | `from` on the visual event's frame |

[Credential-related source omitted]

Install with `npx remotion add @remotion/elevenlabs`. This is the request from Remotion's `elevenLabsTranscriptToCaptions()` docs (https://www.remotion.dev/docs/elevenlabs); the raw REST call keeps the JSON in the shape the function expects

const response = await fetch("https://api.elevenlabs.io/v1/speech-to-text", {

[Credential-related source omitted]

Follow official `display-captions.md` exactly: 1. Fetch the element source (`https://www.remotion.dev/elements/captions/<element>.md`) and copy its `.tsx` file into the project.

1. Fetch the element source (`https://www.remotion.dev/elements/captions/<element>.md`) and copy its `.tsx` file into the project.

Set this up only when the user asks for automated renders: day-to-day work previews in Studio, and an MP4 is rendered only on request. The `npx remotion render` / `npx remotion still` commands and flags follow the official `remotion-render`

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

aspect_ratio: "16:9"         # the default is 1:1, so always pass it   resolution: "2K"             # default 1K; use 2K+ for 1080p compositions   image_urls: ["https://..."]  # optional style/subject references, up to 8, public URLs only

image_urls: ["https://..."]  # optional style/subject references, up to 8, public URLs only

- **remotion-media** `generate_image` writes `public/<output_name>.png`. Put the type in the name: `output_name: "img-hero"` becomes `staticFile("img-hero.png")`. - **Replicate** (the prediction's `output` URL, after `get_predictions` repor

- **Replicate** (the prediction's `output` URL, after `get_predictions` reports `succeeded`) and **Pexels** return URLs. Download them yourself: `mkdir -p public/images && curl -L -o public/images/img-hero.png "<url>"`, then `staticFile("im

[Credential-related source omitted]

mkdir -p public/footage curl -L -o public/footage/city-skyline.mp4 "VIDEO_FILE_LINK"

curl -L -o public/footage/city-skyline.mp4 "VIDEO_FILE_LINK"

Captions follow the official `remotion-captions` skill: word-level transcription with `@remotion/whisper-webgpu` when a GPU is available. Without one, fall back to ElevenLabs Scribe (official `speech-to-text` skill, `scribe_v2` with word ti

mkdir -p public/images public/footage curl -L -o public/images/img-hero.png "<output URL>"

curl -L -o public/images/img-hero.png "<output URL>"

curl -L -o public/footage/clip-city.mp4 "<output URL>"

- `Prefer: "wait"` on the create call blocks for up to 60 s. Use it only for fast image models, never for video. - Start independent predictions in parallel (one per scene), then poll them together. - `download_files` is for files in Replic

- Start independent predictions in parallel (one per scene), then poll them together. - `download_files` is for files in Replicate's Files API (ones uploaded with `create_files`), not for prediction outputs. Outputs are plain delivery URLs:

- `download_files` is for files in Replicate's Files API (ones uploaded with `create_files`), not for prediction outputs. Outputs are plain delivery URLs: `curl -L` them.

1. **Hosted SFX first, for generic sounds.** The official `remotion-markup` skill (sfx.md) lists free files at `https://remotion.media/*.wav`: whoosh, whip, page-turn, switch, mouse-click, shutter-modern, shutter-old, ding, record-scratch, 

| Need | Source | |---|---| | Plain whoosh, click, switch, ding, camera shutter, page turn | Hosted (`https://remotion.media/whoosh.wav`, `mouse-click.wav`, `switch.wav`, `ding.wav`, `shutter-modern.wav`, `page-turn.wav`) |

|---|---| | Plain whoosh, click, switch, ding, camera shutter, page turn | Hosted (`https://remotion.media/whoosh.wav`, `mouse-click.wav`, `switch.wav`, `ding.wav`, `shutter-modern.wav`, `page-turn.wav`) | | Meme sounds for short-form (vine

| Plain whoosh, click, switch, ding, camera shutter, page turn | Hosted (`https://remotion.media/whoosh.wav`, `mouse-click.wav`, `switch.wav`, `ding.wav`, `shutter-modern.wav`, `page-turn.wav`) |

// Hosted whoosh on the slide transition, generated impact just before the cut <Audio src="https://remotion.media/whoosh.wav" from={3 * fps} volume={0.6} />

<Audio src="https://remotion.media/whoosh.wav" from={3 * fps} volume={0.6} />

# Download video (use the rendition's file link from pexels_get_video) curl -L -o public/footage/scene-city.mp4 "https://videos.pexels.com/..."

curl -L -o public/footage/scene-city.mp4 "https://videos.pexels.com/..."

mkdir -p public/images curl -L -o public/images/hero-bg.jpg "https://images.pexels.com/..."

curl -L -o public/images/hero-bg.jpg "https://images.pexels.com/..."

**Always pass `modelName: "pegasus1.5"`.** On 1.0.0, `sync-analyse-video` falls back to an older model the API no longer accepts, and `videoId` inputs only work with that old model, so use `assetId` (or a direct `videoUrl`) instead. For any

generate_video_from_image:   prompt: "<how the image should move>"   image_urls: ["https://..."]       # 1 to animate, 2 to transition; public URLs only

prompt: "<how the image should move>"   image_urls: ["https://..."]       # 1 to animate, 2 to transition; public URLs only   output_name: "clip-hero-move"

image_urls: ["https://..."]       # 1 to animate, 2 to transition; public URLs only

prompt: "gentle push in, clouds slowly moving, subtle parallax"     # duration, aspect ratio, mode: take names and allowed values from the schema # then get_predictions until succeeded, and curl -L the output into public/footage/

# then get_predictions until succeeded, and curl -L the output into public/footage/

- **remotion-media** (`generate_video_from_text`, `generate_video_from_image`) writes `public/<output_name>.mp4`. Put the type in the name: `output_name: "clip-city-aerial"` becomes `staticFile("clip-city-aerial.mp4")`. - **Replicate** retu

- **Replicate** returns the URL in the prediction's `output` once `get_predictions` reports `succeeded`. Download it within the hour: `mkdir -p public/footage && curl -L -o public/footage/clip-city-aerial.mp4 "<output_url>"`, then `staticFi

[Credential-related source omitted]

[Credential-related source omitted]

- **TwelveLabs** has no Agent Skills; point to https://docs.twelvelabs.io.

Official servers from other providers that pair well with Remotion videos. Mention them only if the user asks for more models: - **fal**: official MCP server with hundreds of image and video models (https://fal.ai/docs/documentation/setting

- **fal**: official MCP server with hundreds of image and video models (https://fal.ai/docs/documentation/setting-up/mcp)

- **Runway**: official MCP server for Runway video models (https://github.com/runwayml/runway-api-mcp-server)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

**How to get it:** the quickest way is `npx -y @elevenlabs/cli@latest onboard init` from the project folder: it signs you in and stores a key for the project. By hand: https://elevenlabs.io → API keys page → create a key.

[Credential-related source omitted]

[Credential-related source omitted]

**How to get it:** https://kie.ai → create an account → copy the API key from the dashboard.

[Credential-related source omitted]

[Credential-related source omitted]

**How to get it:** https://www.twelvelabs.io → sign up → dashboard → API keys.

[Credential-related source omitted]

[Credential-related source omitted]

**Pricing:** free tier with limited storage; paid plans for more. Agent docs: https://docs.twelvelabs.io.

**How to get it:** https://www.pexels.com/api/ → "Your API Key" → sign in. The key is shown immediately.

[Credential-related source omitted]

[Credential-related source omitted]

**How to get it:** https://replicate.com → account → API tokens.

[Credential-related source omitted]

[Credential-related source omitted]

- Restart your agent after setting environment variables - For `npx` servers (remotion-media, TwelveLabs, Pexels, Replicate): Node.js 22.18 or newer - For the deprecated ElevenLabs MCP server (`uvx`): install uv (`curl -LsSf https://astral.

- For the deprecated ElevenLabs MCP server (`uvx`): install uv (`curl -LsSf https://astral.sh/uv/install.sh | sh`)

- For the deprecated ElevenLabs MCP server (`uvx`): install uv (`curl -LsSf https://astral.sh/uv/install.sh | sh`)

**Recommended:** - Python 3 + `uv`/`uvx` (only for the optional, deprecated ElevenLabs MCP server)   - If missing, tell user: `curl -LsSf https://astral.sh/uv/install.sh | sh`

- Python 3 + `uv`/`uvx` (only for the optional, deprecated ElevenLabs MCP server)   - If missing, tell user: `curl -LsSf https://astral.sh/uv/install.sh | sh` - Nothing extra for captions: word-level captions use the official `remotion-capt

- If missing, tell user: `curl -LsSf https://astral.sh/uv/install.sh | sh`

- If missing, tell user: `curl -LsSf https://astral.sh/uv/install.sh | sh`

### 4a. ElevenLabs API Key (REQUIRED: voice, sound effects, music) - **What:** Voiceovers, sound effects and music (Eleven Music, cleared for commercial use on paid plans), plus transcription for word-level captions when there's no GPU. Gen

- **What:** Voiceovers, sound effects and music (Eleven Music, cleared for commercial use on paid plans), plus transcription for word-level captions when there's no GPU. Generated through the official ElevenLabs skills. - **Signup:** https:

- **Signup:** https://elevenlabs.io (free tier available; a paid plan is needed for commercial music). Or run `npx -y @elevenlabs/cli@latest onboard init`, which signs in and stores a key for the project.

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

### 4b. KIE API Key (REQUIRED: AI images and clips) - **What:** AI image generation (Nano Banana Pro) and AI video clips (Veo 3.1) through the remotion-media server. It can also make cheap audio drafts, but KIE resells those models without 

- **What:** AI image generation (Nano Banana Pro) and AI video clips (Veo 3.1) through the remotion-media server. It can also make cheap audio drafts, but KIE resells those models without their owners' licence, so final commercial audio sho

- **Signup:** https://kie.ai (create an account, get an API key from the dashboard)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

### 4c. TwelveLabs API Key (REQUIRED: video understanding and AI review) - **What:** Lets your agent watch video: analyse footage, find moments, and review your rendered video with actionable fixes. - **Signup:** https://www.twelvelabs.io (

- **What:** Lets your agent watch video: analyse footage, find moments, and review your rendered video with actionable fixes. - **Signup:** https://www.twelvelabs.io (free tier available) - **Variable:** `TWELVELABS_API_KEY`

- **Signup:** https://www.twelvelabs.io (free tier available)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

### 4d. Pexels API Key (RECOMMENDED: free stock footage) - **What:** Free stock photos and videos. Search by keyword and download HD/4K clips into your project. - **Signup:** https://www.pexels.com/api/ (free, instant)

- **What:** Free stock photos and videos. Search by keyword and download HD/4K clips into your project. - **Signup:** https://www.pexels.com/api/ (free, instant) - **Variable:** `PEXELS_API_KEY`

- **Signup:** https://www.pexels.com/api/ (free, instant)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

### 4e. Replicate API Token (OPTIONAL: specific image/video models) - **What:** Hundreds of models (FLUX.2, Kling, Seedance, Wan and more) when you want a specific model, or as a fallback when a KIE clip fails. - **Signup:** https://replica

- **What:** Hundreds of models (FLUX.2, Kling, Seedance, Wan and more) when you want a specific model, or as a fallback when a KIE clip fails. - **Signup:** https://replicate.com (create an account, get an API token) - **Variable:** `REPLIC

- **Signup:** https://replicate.com (create an account, get an API token)

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]