Media generation: runnable scripts for media generation at scale

.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "runway-api",
-  "description": "Helps users integrate Runway's public API (https://docs.dev.runwayml.com/) into their projects. Analyzes codebase compatibility, guides API key setup, and provides hands-on assistance for video generation, image generation, audio, file uploads, and direct API actions.",
-  "version": "1.2.0",
+  "description": "Video generation at scale. Generate videos, images, and audio with Runway's API — batch ad campaigns, product videos, multishot stories, and creative iteration. Supports seedance2, gen4.5, veo3, Nano, Banana Pro, and more.",
+  "version": "2.0.0",
   "author": {
     "name": "Runway",
     "email": "ops@runwayml.com"
@@ -11,12 +11,13 @@
     "runwayml",
     "video-generation",
     "image-generation",
+    "audio-generation",
     "ai-api",
+    "seedance2",
     "gen4",
     "veo3",
-    "characters",
-    "avatars",
-    "public-api",
+    "media-generation",
+    "batch",
     "runway-api"
   ]
 }

.cursor-plugin/plugin.json

@@ -1,8 +1,8 @@
 {
   "name": "runway-api",
   "displayName": "Runway API",
-  "version": "1.0.1",
-  "description": "Helps users integrate Runway's public API (https://docs.dev.runwayml.com/) into their projects. Analyzes codebase compatibility, guides API key setup, and provides hands-on assistance for video generation, image generation, audio, file uploads, and direct API actions.",
+  "version": "2.0.0",
+  "description": "Video generation at scale. Generate videos, images, and audio with Runway's API — batch ad campaigns, product videos, multishot stories, and creative iteration. Supports seedance2, gen4.5, veo3, Nano, Banana Pro, and more.",
   "author": {
     "name": "Runway",
     "email": "ops@runwayml.com"
@@ -18,18 +18,13 @@
     "image-generation",
     "audio-generation",
     "ai-api",
+    "seedance2",
     "gen4",
     "veo3",
-    "characters",
-    "avatars"
-  ],
-  "category": "ai-tools",
-  "tags": [
-    "api-integration",
     "media-generation",
-    "video",
-    "image",
-    "audio"
+    "batch"
   ],
+  "category": "ai-tools",
+  "tags": ["media-generation", "video", "image", "audio", "api-integration"],
   "skills": "./skills/"
 }

.gitignore

@@ -26,3 +26,6 @@ __pycache__/
 
 # Dependencies
 node_modules/
+
+# Test output files
+test-*

CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## 2.0.0
+
+- **Media generation:** New `rw-generate-video`, `rw-generate-image`, `rw-generate-audio` skills that run Python scripts directly via `uv run` — no SDK setup required
+- **Runnable scripts:** Added `scripts/` directory with `generate_video.py`, `generate_image.py`, `generate_audio.py`, `list_models.py`, `get_task.py`, and shared `runway_helpers.py`
+- **Seedance 2 support:** Added `seedance2` model across all generation scripts and skills (TTV, ITV, VTV, 36 credits/sec)
+- **Plugin metadata:** Updated descriptions and keywords for both Claude and Cursor plugins
+
 ## 1.1.0
 
 - **Breaking change:** Every skill is now named with an `rw-` prefix (skill folder under `skills/`, `name` in each `SKILL.md`, and `+…` invocations). Examples: `setup-api-key` → `rw-setup-api-key`, `integrate-video` → `rw-integrate-video`. Update any documentation, shortcuts, or automation that referenced the previous names or paths.

README.md

@@ -1,37 +1,39 @@
 # Runway API Skills
 
-A set of skills that gives your AI coding agent the knowledge and tools to work with [Runway's public API](https://docs.dev.runwayml.com/api/) — integrate video, image, and audio generation into server-side projects, or manage Runway resources directly from the editor.
+Video generation at scale. Generate videos, images, and audio with Runway's API — batch Ad campaigns, product videos, multishot stories, and creative iteration. Supports seedance2, gen4.5, veo3, Nano, Banana Pro, and more.
 
-Works with [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex](https://openai.com/index/codex/), and other compatible agents.
+Works with [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Cursor](https://cursor.com), [Codex](https://openai.com/index/codex/), and other compatible agents.
 
 ## Two Ways to Use
 
-### 1. Integrate into your app
+### 1. Generate media at scale
 
-Guide your agent through adding Runway capabilities to a server-side project: verify compatibility, set up credentials, write framework-specific routes, and handle edge cases like file uploads and task polling.
+Tell your agent what to create and it handles the rest — calls the API, polls for completion, downloads the result. Generate one asset or orchestrate hundreds.
 
 ```
-Set up Runway video generation in my Next.js app
+Generate a 15-second product video from this image using seedance2
 ```
 
-### 2. Act on your Runway account directly
-
-Let your agent call the Runway API to manage resources — create avatars, update knowledge documents, trigger generations, check credit balance — without writing app code.
+```
+Generate a video for each of these 5 product photos, 9:16 ratio for Instagram Reels
+```
 
 ```
-Create a new avatar called "Support Agent" with a friendly personality
+Create a voiceover for this ad script, then generate a sound effect of applause
 ```
 
+### 2. Integrate into your app
+
+Guide your agent through adding Runway capabilities to a server-side project: verify compatibility, set up credentials, write framework-specific routes, and handle edge cases like file uploads and task polling.
+
 ```
-List my knowledge documents for avatar abc123
+Set up Runway video generation in my Next.js app
 ```
 
 ## Installation
 
 ### Claude Code (community marketplace)
 
-Add the [Anthropic community plugins](https://github.com/anthropics/claude-plugins-community) marketplace, then install this plugin.
-
 ```bash
 claude plugin marketplace add anthropics/claude-plugins-community
 claude plugin install runway-api-skills@claude-community
@@ -50,34 +52,40 @@ Select all the skills with your keyboard (Space to select, arrow keys to navigat
 ## Prerequisites
 
 - A [Runway developer account](https://dev.runwayml.com/) with prepaid credits ($10 minimum)
+- For generation skills: [uv](https://docs.astral.sh/uv/) (Python package runner) and `RUNWAYML_API_SECRET` env var
 - For integration skills: a server-side project — Node.js 18+ or Python 3.8+ with a backend framework
-- For direct API actions: Node.js 20+ (zero dependencies — just the runtime script)
 
 ## Available Skills
 
-### Getting Started
+### Generation (run directly)
 
-| Skill                    | Description                                                                                    |
-| ------------------------ | ---------------------------------------------------------------------------------------------- |
-| `rw-recipe-full-setup`   | End-to-end setup: compatibility check → API key → SDK install → integration code → test        |
-| `rw-check-compatibility` | Analyze your project to verify it can safely call the Runway API server-side                   |
-| `rw-setup-api-key`       | Guide through account creation, SDK installation, and environment variable configuration       |
-| `rw-check-org-details`   | Query your organization's rate limits, credit balance, usage tier, and daily generation counts |
+Generate media assets directly — your agent runs the scripts, polls for completion, and saves the output.
 
-### Direct API Actions
+| Skill                | Description                                                                         |
+| -------------------- | ----------------------------------------------------------------------------------- |
+| `rw-generate-video`  | Generate videos: text-to-video, image-to-video, video-to-video                     |
+| `rw-generate-image`  | Generate images: text-to-image with optional reference images                       |
+| `rw-generate-audio`  | Generate audio: TTS, sound effects, voice isolation, dubbing, voice conversion      |
 
-| Skill               | Description                                                                        |
-| -------------------- | ---------------------------------------------------------------------------------- |
-| `use-runway-api`     | Call any public API endpoint to manage resources, trigger generations, and inspect state |
+### Integration (add to your app)
 
-### Generation (Integration)
+Add Runway generation to your server-side project with framework-specific code.
 
 | Skill                | Description                                                                         |
 | -------------------- | ----------------------------------------------------------------------------------- |
 | `rw-integrate-video` | Text-to-video, image-to-video, video-to-video, and character performance generation |
 | `rw-integrate-image` | Text-to-image generation with optional reference images via `@Tag` syntax           |
 | `rw-integrate-audio` | Text-to-speech, sound effects, voice isolation, dubbing, and speech-to-speech       |
 
+### Getting Started
+
+| Skill                    | Description                                                                                    |
+| ------------------------ | ---------------------------------------------------------------------------------------------- |
+| `rw-recipe-full-setup`   | End-to-end setup: compatibility check → API key → SDK install → integration code → test        |
+| `rw-check-compatibility` | Analyze your project to verify it can safely call the Runway API server-side                   |
+| `rw-setup-api-key`       | Guide through account creation, SDK installation, and environment variable configuration       |
+| `rw-check-org-details`   | Query your organization's rate limits, credit balance, usage tier, and daily generation counts |
+
 ### Characters (Real-Time Avatars)
 
 | Skill                          | Description                                                                                |
@@ -90,6 +98,7 @@ Select all the skills with your keyboard (Space to select, arrow keys to navigat
 
 | Skill                    | Description                                                                                                           |
 | ------------------------ | --------------------------------------------------------------------------------------------------------------------- |
+| `use-runway-api`         | Call any public API endpoint to manage resources, trigger generations, and inspect state                               |
 | `rw-integrate-uploads`   | Upload local files to get `runway://` URIs for use as generation inputs                                               |
 | `rw-api-reference`       | Complete API reference — models, endpoints, costs, rate limits, and error codes                                       |
 | `rw-fetch-api-reference` | Fetch the latest API docs from [docs.dev.runwayml.com/api](https://docs.dev.runwayml.com/api/) as the source of truth |
@@ -98,14 +107,14 @@ Select all the skills with your keyboard (Space to select, arrow keys to navigat
 
 ### Video
 
-| Model                    | Use Case                            | Cost              |
-| ------------------------ | ----------------------------------- | ----------------- |
+| Model                    | Use Case                                 | Cost              |
+| ------------------------ | ---------------------------------------- | ----------------- |
 | `seedance2`              | Reference image and video, long duration | 36 credits/sec    |
-| `gen4.5`                 | High quality, general purpose       | 12 credits/sec    |
-| `gen4_turbo`             | Fast, image-driven (image required) | 5 credits/sec     |
-| `gen4_aleph`             | Video-to-video editing              | 15 credits/sec    |
-| `veo3`                   | Premium quality                     | 40 credits/sec    |
-| `veo3.1` / `veo3.1_fast` | High quality / fast Google models   | 10–40 credits/sec |
+| `gen4.5`                 | High quality, general purpose            | 12 credits/sec    |
+| `gen4_turbo`             | Fast, image-driven (image required)      | 5 credits/sec     |
+| `gen4_aleph`             | Video-to-video editing                   | 15 credits/sec    |
+| `veo3`                   | Premium quality                          | 40 credits/sec    |
+| `veo3.1` / `veo3.1_fast` | High quality / fast Google models       | 10–40 credits/sec |
 
 ### Image
 
@@ -133,34 +142,28 @@ Select all the skills with your keyboard (Space to select, arrow keys to navigat
 
 ## Quick Start
 
-### Integrate into a project
+### Generate media directly
 
 ```
-Set up Runway video generation in my Next.js app
+Generate a 10-second video of a sunset over the ocean
 ```
 
-Or go step by step:
-
 ```
-Check if my project is compatible with the Runway API
+Generate an image of a red door in a white wall
 ```
 
 ```
-Help me set up my Runway API key
+Create a voiceover saying "Welcome to our store" and save it as welcome.mp3
 ```
 
-```
-Add an endpoint to generate videos from text prompts
-```
-
-### Direct API actions
+### Integrate into a project
 
 ```
-List all my avatars
+Set up Runway video generation in my Next.js app
 ```
 
 ```
-Using the Runway API, generate an image of a red door in a white wall and tell me which model you used
+Add an endpoint to generate videos from text prompts
 ```
 
 ## Supported Frameworks

scripts/generate_audio.py

@@ -0,0 +1,131 @@
+# /// script
+# requires-python = ">=3.10"
+# dependencies = ["requests"]
+# ///
+
+"""Generate audio using the Runway API (TTS, sound effects, voice isolation, dubbing)."""
+
+import argparse
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+from runway_helpers import (
+    get_api_key,
+    api_post,
+    poll_task,
+    download_file,
+    ensure_url,
+    output_path,
+)
+
+AUDIO_TYPES = {
+    "tts": {
+        "endpoint": "/v1/text_to_speech",
+        "model": "eleven_multilingual_v2",
+        "description": "Text to speech",
+    },
+    "sfx": {
+        "endpoint": "/v1/sound_effect",
+        "model": "eleven_text_to_sound_v2",
+        "description": "Sound effect generation",
+    },
+    "isolate": {
+        "endpoint": "/v1/voice_isolation",
+        "model": "eleven_voice_isolation",
+        "description": "Isolate voice from audio",
+    },
+    "dub": {
+        "endpoint": "/v1/voice_dubbing",
+        "model": "eleven_voice_dubbing",
+        "description": "Dub to another language",
+    },
+    "sts": {
+        "endpoint": "/v1/speech_to_speech",
+        "model": "eleven_multilingual_sts_v2",
+        "description": "Voice conversion",
+    },
+}
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Generate audio with the Runway API")
+    parser.add_argument("--filename", required=True, help="Output filename (e.g. output.mp3)")
+    parser.add_argument(
+        "--type",
+        required=True,
+        choices=list(AUDIO_TYPES.keys()),
+        help="Audio type: tts, sfx, isolate, dub, sts",
+    )
+    parser.add_argument("--text", help="Text input (required for tts and sfx)")
+    parser.add_argument("--audio-url", help="Audio URL or local path (for isolate, dub, sts)")
+    parser.add_argument("--voice-id", help="Voice ID (for tts and sts)")
+    parser.add_argument("--target-language", help="Target language code (for dub, e.g. 'es')")
+    parser.add_argument("--output-dir", help="Output directory (default: cwd)")
+    parser.add_argument("--api-key", help="Runway API key (or set RUNWAYML_API_SECRET)")
+    args = parser.parse_args()
+
+    api_key = get_api_key(args.api_key)
+    audio_type = AUDIO_TYPES[args.type]
+    endpoint = audio_type["endpoint"]
+    model = audio_type["model"]
+
+    body = {"model": model}
+
+    if args.type == "tts":
+        if not args.text:
+            print("Error: --text is required for tts.", file=sys.stderr)
+            sys.exit(1)
+        body["promptText"] = args.text
+        body["voice"] = {"type": "runway-preset", "presetId": args.voice_id or "Maya"}
+
+    elif args.type == "sfx":
+        if not args.text:
+            print("Error: --text is required for sfx.", file=sys.stderr)
+            sys.exit(1)
+        body["promptText"] = args.text
+
+    elif args.type == "isolate":
+        if not args.audio_url:
+            print("Error: --audio-url is required for isolate.", file=sys.stderr)
+            sys.exit(1)
+        body["audioUri"] = ensure_url(args.audio_url, api_key)
+
+    elif args.type == "sts":
+        if not args.audio_url:
+            print("Error: --audio-url is required for sts.", file=sys.stderr)
+            sys.exit(1)
+        audio_uri = ensure_url(args.audio_url, api_key)
+        body["media"] = {"type": "audio", "uri": audio_uri}
+        body["voice"] = {"type": "runway-preset", "presetId": args.voice_id or "Maya"}
+
+    elif args.type == "dub":
+        if not args.audio_url:
+            print("Error: --audio-url is required for dub.", file=sys.stderr)
+            sys.exit(1)
+        if not args.target_language:
+            print("Error: --target-language is required for dub.", file=sys.stderr)
+            sys.exit(1)
+        body["audioUri"] = ensure_url(args.audio_url, api_key)
+        body["targetLang"] = args.target_language
+
+    print(f"Generating audio ({args.type}) with {model}...", file=sys.stderr)
+    task = api_post(api_key, endpoint, body)
+    task_id = task.get("id")
+    print(f"Task created: {task_id}", file=sys.stderr)
+
+    result = poll_task(api_key, task_id)
+    urls = result.get("output", [])
+
+    if not urls:
+        print("Error: No output URLs in result.", file=sys.stderr)
+        sys.exit(1)
+
+    out = output_path(args.filename, args.output_dir)
+    path = download_file(urls[0], out)
+    print(path)
+    print(f"Saved: {path}", file=sys.stderr)
+
+
+if __name__ == "__main__":
+    main()