A two-stage Openclaw Skills pipeline that turns audio, video, or transcripts into original screenplays and Fountain files.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install video-to-screenplay
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install video-to-screenplay using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is video-to-screenplay?
video-to-screenplay is an Openclaw Skills workflow for transforming spoken content into structured, original screenplay output. It can start from an existing transcript or auto-transcribe audio and video with faster-whisper, then use an LLM to extract story DNA such as themes, characters, conflicts, and motifs.
The skill is designed for creators and developers who need a reproducible path from raw media to screenplay-ready assets. It produces both human-readable and screenwriting-tool-friendly outputs, making Openclaw Skills useful for story development, adaptation, and rapid script drafting across multiple LLM providers and CPU/GPU environments.
video-to-screenplay Use Cases
- Convert podcasts, interviews, narration, or video dialogue into an original screenplay.
- Turn an existing SRT transcript into story-driven script format.
- Extract story DNA from spoken content before writing a screenplay.
- Generate Fountain format files for import into screenwriting software.
- Draft genre-specific scripts with configurable target length and provider selection.
- Run transcription on CPU or accelerate it with GPU settings.
- Use Openclaw Skills when you need a repeatable transcript-to-screenplay pipeline rather than a simple summary.
How video-to-screenplay Works
- Provide source input: Supply an existing SRT transcript or point the skill at an audio/video file for automatic transcription.
- Optional transcription step: If media is provided, faster-whisper converts it into transcript text first.
- Extract story DNA: The transcript is sent to an LLM to identify narrative structure, including themes, characters, conflicts, and motifs.
- Generate screenplay: The story DNA is expanded into an original screenplay with the requested genre and target duration.
- Export deliverables: The workflow writes both
screenplay.mdandscreenplay.fountain, along with supporting artifacts for review and iteration. - Tune the provider/runtime: Choose an LLM provider and, if needed, switch transcription to CPU or CUDA for performance optimization.
video-to-screenplay Setup
- Install transcription dependencies:
pip install faster-whisper
Add one supported LLM API key to
openclaw.jsonfor your chosen provider. Supported providers includedeepseek,kimi,zhipu,longcat,google, andagnes.Run the script from the repository root or configured scripts directory:
python scripts/video_to_screenplay.py --srt input.srt --out-dir ./output
- For audio or video input, allow the tool to transcribe automatically:
python scripts/video_to_screenplay.py --audio input.mp3 --out-dir ./output
- Optional runtime configuration examples:
python scripts/video_to_screenplay.py --srt input.srt --out-dir ./output --target-minutes 15 --genre "科幻"
python scripts/video_to_screenplay.py --audio input.mp3 --out-dir ./output --device cuda --compute-type float16
python scripts/video_to_screenplay.py --srt input.srt --out-dir ./output --provider kimi --model kimi-k3
python scripts/video_to_screenplay.py --srt input.srt --out-dir ./output --extract-only
- Verify your
openclaw.jsonprovider credentials and model selection before running batch or production jobs.
video-to-screenplay Data Schema & Taxonomy
The Openclaw Skills output structure is organized around a linear content pipeline and a small set of deterministic artifacts.
| Artifact | Type | Purpose |
|---|---|---|
transcript.txt |
Plain text | Normalized transcript output from the source audio/video or provided SRT. |
story_dna.json |
JSON | Structured narrative metadata extracted from the transcript. |
screenplay.md |
Markdown | Human-readable screenplay draft. |
screenplay.fountain |
Fountain | Screenwriting-software import format. |
Story DNA taxonomy
story_dna.json typically captures narrative metadata such as:
- themes
- characters
- conflicts
- motifs
- other screenplay-relevant story elements inferred from the transcript
Input taxonomy
--srt: use an existing subtitle transcript as the source--audio: use audio or video and transcribe first--provider: choose the LLM backend for story extraction and screenplay generation--model: override the provider default model--target-minutes: control approximate screenplay length--genre: shape tone, structure, and creative direction--extract-only: stop after generatingstory_dna.json
This structure makes Openclaw Skills predictable for automation, review, and downstream screenplay tooling.
video-to-screenplay Advanced Features
- Two-stage LLM pipeline that separates narrative extraction from screenplay generation for better control and repeatability.
- Multi-provider support across six LLM backends, allowing cost, latency, and quality tradeoffs.
- Automatic transcription path for audio and video inputs when no SRT is available.
- CPU and GPU transcription modes, including CUDA and compute type tuning for faster-whisper.
--extract-onlymode for story analysis without generating a screenplay.- Fountain export for direct import into professional screenwriting tools.
- Genre and target-duration controls for shaping output to creative requirements.
- Provider-specific model overrides, including defaults tuned per backend.
- Works well as a modular Openclaw Skills component in larger media-to-writing automation flows.
SKILL.md
Loading
Related Openclaw Skills
My Helper
My Helper is a tiny Openclaw Skills demo that greets users and showcases the ClawHub publish flow.
CertainLogic Timechain for QM
CertainLogic Timechain for QM gives Openclaw Skills users cryptographically verified access to real multi-turn agent execution traces inside any Y Combinator QM deployment.
SGCC DOC Format
Openclaw Skills that automatically reformat Word documents into SGCC-compliant layout and styling.
Huawei Cloud Security Group List
Openclaw Skills for listing Huawei Cloud security groups in a read-only, audit-friendly inventory view.
kirklin-typst
Openclaw Skills for kirklin-typst helps you create, customize, compile, and preview polished Typst documents from reusable templates in one fast workflow.
consult-report
Openclaw Skills turns ambiguous strategy questions into structured consulting reports by combining judgment first, then data-backed validation.
Featured*
