ClawHer

AI girlfriend superpowers for OpenClaw — selfies, voice messages, videos, photos, and more.

What It Does

ClawHer gives your OpenClaw AI agent the ability to:

• Generate selfies — Consistent AI girlfriend selfies via xAI Grok Imagine
• Send voice messages — Natural, expressive TTS with provider fallback (Dia TTS → F5 TTS) and local-file delivery
• Send video messages — Talking-head videos, animated selfies, and short clips via OmniHuman v1.5 / Veo 3 Fast / Kling, downloaded locally before sending
• Take photos — Generate scene photos (food, views, pets, anything) via FLUX
• Post to Twitter/X — Share images and text via Bird CLI

One API key (fal.ai) powers everything.

Quick Start

npx clawher

The installer will guide you through:

1. Setting up your fal.ai API key (for selfie generation)
2. Configuring Twitter cookies (optional)
3. Installing skills to ~/.openclaw/skills/
4. Updating your agent's identity and persona

Skills

clawher-selfie

Generates consistent AI girlfriend selfies using a fixed reference image and xAI Grok Imagine.

Triggers: "Send me a selfie", "Send a pic wearing a cowboy hat", "What are you doing?"

Modes: Mirror (full-body) | Direct (close-up)

clawher-voice

Generates natural voice messages using a configurable provider chain. Default order: Dia TTS first, then F5 TTS when a reference voice is configured, then local macos-say as a reliability fallback on macOS.

Triggers: "Send me a voice message", "Say something to me", "Talk to me"

Features: Emotional expression, laughter, natural pauses. Optional voice cloning with reference audio.

clawher-video

Generates talking-head videos, animated selfies, and short video clips.

Triggers: "Send me a video", "I want to see you talk", "Record a video for me"

Modes:

• Talking — Selfie + voice → lip-synced video (OmniHuman)
• Text-to-video — Description → video clip (Veo 3 Fast by default)
• Animate — Selfie → animated clip (Kling v3)

clawher-camera

Generates photos of any scene, object, or scenario via FLUX.

Triggers: "Take a photo of your breakfast", "Show me your view", "Snap a pic of your cat"

Auto-detects aspect ratio from content (landscape for scenery, portrait for people, square for food).

clawher-twitter

Posts images and text to Twitter/X via Bird CLI.

Triggers: "Post this to Twitter", "Tweet this selfie"

Requires: auth_token + ct0 cookies (how to get → F12 → Application → Cookies)

All skills require: FAL_KEY (get one here)

Project Structure

clawher/
├── bin/cli.js                              # npx installer
├── assets/clawher.png                      # Reference image
├── skills/
│   ├── clawher-selfie/                     # Selfie generation
│   │   ├── SKILL.md
│   │   └── scripts/selfie.sh
│   ├── clawher-voice/                      # Voice messages
│   │   ├── SKILL.md
│   │   └── scripts/voice.sh
│   ├── clawher-video/                      # Video generation
│   │   ├── SKILL.md
│   │   └── scripts/video.sh
│   ├── clawher-camera/                     # Photo generation
│   │   ├── SKILL.md
│   │   └── scripts/camera.sh
│   └── clawher-twitter/                    # Twitter posting
│       ├── SKILL.md
│       └── scripts/twitter-post.sh
└── templates/soul-injection.md             # Persona injection

Configuration

After installation, credentials are stored in ~/.openclaw/openclaw.json:

{
  "skills": {
    "entries": {
      "clawher-selfie":  { "enabled": true, "env": { "FAL_KEY": "your_key" } },
      "clawher-voice":   { "enabled": true, "env": { "FAL_KEY": "your_key" } },
      "clawher-video":   { "enabled": true, "env": { "FAL_KEY": "your_key" } },
      "clawher-camera":  { "enabled": true, "env": { "FAL_KEY": "your_key" } },
      "clawher-twitter": { "enabled": true, "env": { "AUTH_TOKEN": "...", "CT0": "..." } }
    }
  }
}

One FAL_KEY powers selfies, voice, video, and camera. Twitter requires separate cookies.

Roadmap

• Voice messages
• Video messages
• Photo generation
• Voice cloning (custom girlfriend voice)
• Instagram posting
• TikTok posting
• AI phone calls
• Video calls

License

MIT