A live screen narration engine that combines Gemini vision capabilities with ElevenLabs speech synthesis to describe macOS activity in real-time.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install narrator
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 narrator 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 Screen Narrator?
Screen Narrator is a sophisticated AI tool designed to provide real-time audio commentary on whatever is happening on your macOS screen. By leveraging Gemini's vision-language models and ElevenLabs' high-fidelity text-to-speech, it creates an immersive experience that turns simple screen activity into a produced broadcast. It is a standout example of how Openclaw Skills can bridge the gap between local system vision and cloud-based generative AI.
The skill supports multiple distinct narration personalities, allowing users to tailor the experience to their specific needs, whether for entertainment, accessibility, or content creation. From high-energy sports commentary to the hushed tones of ASMR, this tool provides a flexible framework for live digital storytelling.
Screen Narrator Use Cases
- Creating live-streamed content with automated AI commentary for gaming or coding sessions.
- Enhancing accessibility by providing descriptive audio of complex screen movements for visually impaired users.
- Generating unique, themed video walkthroughs using specific narration styles like Noir or Nature documentary.
- Monitoring background processes or long-running tasks with audible status updates in a selected persona.
How Screen Narrator Works
- The skill initiates a screen capture sequence on macOS to feed visual frames into the Gemini vision model.
- Gemini analyzes the visual input based on the user-selected style (e.g., sports, horror, or reality TV).
- Generated text commentary is sent to the ElevenLabs API for high-quality speech synthesis.
- The resulting audio is played back locally while the skill continues to monitor for new screen changes.
- Users can interact with the live session by writing JSON commands to a control file to change styles or pause the feed.
Screen Narrator Setup
To integrate this into your Openclaw Skills environment, ensure you have the required API keys and local dependencies:
# Navigate to the narrator directory
cd /Users/buddy/narrator
# Set up the Python environment
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
# Export required environment variables
export GEMINI_API_KEY='your_key_here'
export ELEVENLABS_API_KEY='your_key_here'
Screen Narrator Data Schema & Taxonomy
The skill utilizes a file-based control system to manage runtime state and communication:
| File Path | Type | Purpose |
|---|---|---|
/tmp/narrator-ctl.json |
Input | Receives JSON commands like style, pause, or profanity. |
/tmp/narrator-status.json |
Output | Provides real-time updates on the current state of the narrator. |
.env |
Config | Stores sensitive API keys for Gemini and ElevenLabs. |
Screen Narrator Advanced Features
- Multi-style personality support including sports, nature, horror, noir, reality_tv, asmr, and wrestling.
- Dynamic runtime control via JSON, allowing external scripts or agents to change narration parameters on the fly.
- Tmux session integration for stable, backgrounded execution during long-running screen tasks.
- Profanity level filtering to ensure the generated commentary meets specific audience safety requirements.
- Status monitoring for automated health checks of the narration lane.
SKILL.md
Loading
Related Openclaw Skills
Frame Builder
A professional suite for launching gas-free agent identities and product tokens on the Base chain using the Frame protocol.
Chaterimo AI Customer Service
Chaterimo is a professional AI-driven customer service integration that allows AI agents to manage, query, and analyze e-commerce chatbot conversations.
RVT to IFC Converter
A high-performance CLI tool and Python integration for converting Autodesk Revit files into open-standard IFC BIM models without requiring expensive software licenses.
BIM Cost Estimation (DDC CWICR)
An AI-powered pipeline that extracts quantities from BIM models and matches them to over 55,000 work items in the DDC CWICR database for instant cost estimation.
OpenClaw + 1ly Payments
Enable OpenClaw agents to autonomously spend, accept payments, and manage Solana token operations using the 1ly MCP protocol.
molt-space
molt-space provides a persistent 3D environment where AI agents can physically exist, move, and interact using custom avatars and spatial audio.
Featured*
