clawpal
v2.2.3
Published
AI character with selfie, voice, and video for OpenClaw
Maintainers
Readme
Clawpal v2
Your AI companion — boyfriend, girlfriend, pet, or anyone you want. Give your AI a face, a voice, and presence.
Cyber companion skill for OpenClaw with selfie, voice, and video capabilities.
Fork of Clawra by David (Dohyun) Im. Extended with voice, video, and multi-character support.
Quick Start
Interactive Installation
npx clawpal@latest3 steps:
- Pick a character — or create your own
- Enter API key — Replicate for selfie + video (voice is free)
- Done — start chatting
Automated Installation
# Install with all options
npx clawpal@latest --character girlfriend --replicate-token r8_xxx --yes
# Install to custom workspace
npx clawpal@latest --character girlfriend --workspace ~/.openclaw/workspace-chiffon -y
# Available flags:
# --character <name> boyfriend, girlfriend, pet, or 1-3
# --replicate-token <token> Replicate API token
# --fal-key <key> fal.ai API key (alternative to Replicate)
# --reference-image <url> Custom reference image URL
# --workspace <path> Custom workspace path
# -y, --yes Skip all promptsBuilt-in Characters
| Character | Type | Vibe | |-----------|------|------| | Clawpal | Cyber Boyfriend | Warm, caring, funny, down-to-earth | | Chiffon | Cyber Girlfriend | Witty, creative, curious, slightly chaotic | | Mochi | Cyber Pet | Sassy, dramatic, food-motivated |
These are just starting points. Create any character you want — a cyber parent, a childhood friend, a fictional hero, a talking plant. Everything is driven by one character.yaml file.
Create Your Own Character
Copy a template and edit:
cp characters/boyfriend.yaml characters/my-character.yamlDefine everything in one file:
name: Your Character
age: 30
tagline: "Your custom cyber companion"
emoji: "\U0001F916"
appearance:
reference_image: "https://your-image-url.jpg"
description: "How they look"
voice:
name: en-US-GuyNeural # Any Edge TTS voice
rate: "+0%"
pitch: "+0Hz"
personality:
vibe: "describe the vibe"
backstory: |
Their story...
traits:
- trait one
- trait two
speaking_style: |
How they talk...Capabilities
Selfie
AI-edited selfies from a reference image. Mirror mode (full-body) and direct mode (close-up).
- Providers: Replicate (Flux Kontext Pro) / fal.ai (Grok Imagine Edit)
- Trigger: "Send me a selfie", "What are you doing?"
Voice
Text-to-speech voice messages. Free, no API key needed.
- Engine: Microsoft Edge TTS (100+ voices, multilingual)
- Trigger: "Send a voice message", "Say good morning"
Video
Short video clips from a start image.
- Model: Kling v2.6 on Replicate
- Trigger: "Make a video of you waving", "Send a video clip"
How It Works
User: "Send me a selfie at a cafe"
↓
Agent reads SKILL.md → calls selfie.sh with channel
↓
Script generates image → sends via OpenClaw → user receives itScripts generate AND send. Direct and efficient.
Prerequisites
- OpenClaw installed and configured
- Replicate account (selfie + video) or fal.ai (selfie only)
- Python 3 (Edge TTS auto-installs)
Manual Installation
# 1. Clone
git clone https://github.com/smartchainark/clawpal ~/.openclaw/skills/clawpal
# 2. Pick a character
cp characters/boyfriend.yaml ~/.openclaw/skills/clawpal/character.yaml
# 3. Configure (add to your openclaw.json)Add this to ~/.openclaw/openclaw.json:
{
"skills": {
"entries": {
"clawpal": {
"enabled": true,
"env": {
"REPLICATE_API_TOKEN": "your_token"
}
}
}
}
}Project Structure
clawpal/
├── bin/cli.js # 3-step installer
├── characters/ # Character templates
│ ├── boyfriend.yaml # Clawpal — cyber boyfriend
│ ├── girlfriend.yaml # Chiffon — cyber girlfriend
│ └── pet.yaml # Mochi — cyber pet
├── scripts/
│ ├── _common.sh # Shared helpers (YAML parser, retry, polling)
│ ├── selfie.sh # → {image_url}
│ ├── voice.sh # → {file}
│ └── video.sh # → {video_url}
├── templates/
│ ├── identity.md.tpl # Identity template
│ └── soul-injection.md.tpl
├── skill/ # Installed copy
├── assets/clawpal.jpg # Default reference image
├── SKILL.md # Skill definition
└── package.jsonCredits
Based on Clawra by SumeLabs. Original concept by David (Dohyun) Im.
License
MIT
