claw3d 0.1.0

CLI for converting sketches/images to 3D models and printing via Moonraker (OpenClaw skill)

claw3d

Open-source OpenClaw skill for converting sketches/images to 3D models and printing via Moonraker. Powers image-to-3D, sketch-to-print, and text-to-3D workflows.

---

What It Does

1. Convert — Sketch/image or text prompt → 3D model (GLB) via FAL.ai (FLUX + Tripo)
2. Search — Find 3D models on Thingiverse (claw3d search "dragon")
3. Scale — Resize 3D model
4. Slice — 3D model → G-code via CuraEngine
5. Print — Upload G-code to Moonraker and start print

→ Full workflow map & architecture — coordinate systems, all flows, improvement roadmap

Install

From PyPI (recommended):

pip install claw3d
# or
uv tool install claw3d

From source (development):

cd claw3d
pip install -e .

The claw3d command is available after install.

Quick Start

# 1. Set your FAL API key
export FAL_API_KEY=your-key

# 2. Convert a sketch or prompt to 3D
claw3d convert --image sketch.png --prompt "a toy robot" --output model.glb          # Sketch → Tripo
claw3d convert --prompt "a blue rocket figurine" --output model.glb                 # Text → image → Tripo
claw3d convert --image multiview.png --multiview --output model.glb                 # Multiview quad → Gemini → Hunyuan
claw3d convert --edit-3d model.glb --prompt "make it blue" --output edited.glb      # 3D-to-3D: render → enhance → Hunyuan
claw3d convert --image sketch.png --no-enhance --output model.glb                   # Direct to Tripo (no enhance)

# Per-step model overrides (e.g. --sketch-enhance-model flux2_pro_edit)
claw3d convert --help  # lists --*-model flags

# 3. Scale (optional)
claw3d scale --input model.glb --output scaled.glb --scale 0.5

# 4. Slice (requires CuraEngine API)
# Or search for models: claw3d search "dragon" --limit 5 (needs THINGIVERSE_ACCESS_TOKEN)
export CLAW3D_SLICER_URL=http://localhost:8000
claw3d slice --input scaled.glb --profile <profile_id> --output model.gcode
# Or GLB + 3MF: claw3d slice -i model.glb --profile-from-3mf settings.3mf -o model.gcode
# Run `claw3d profile list` first to get profile_id; if multiple, ask which one.

# 5. Print (requires Moonraker on LAN)
claw3d print --gcode model.gcode [--printer <id>]
# Run `claw3d printers` first; if multiple printers, ask which one.

Slicing API

claw3d uses the CuraEngine slicing API (curaengine-slicer-api). It lives in the sibling folder ../curaengine-slicer-api. With Docker Compose, the slicer runs as a companion service. Standalone:

# Run slicer (sibling folder)
cd ../curaengine-slicer-api
docker compose up -d

Then create a slicing profile via claw3d profile create --from-3mf settings.3mf --name my_printer or POST /profiles/from-3mf.

OpenClaw Skill

I have OpenClaw + Telegram. What's next? Run the setup script — it asks which skills you want:

cd claw3d
./scripts/setup-openclaw-docker.sh ../openclaw-base

• 3d-ai-forger — Convert images/text to 3D. Prompts for FAL_API_KEY.
• 3d-printing — Slice and print. Choose local or cloud slicer.
• 3d-directory — Search Thingiverse. Prompts for THINGIVERSE_ACCESS_TOKEN.

The installer prompts for API keys (FAL, Thingiverse) and writes them to .env. Use --force-recreate when starting so the gateway uses the freshly built image.

Test from scratch: docs/TEST_FROM_SCRATCH.md — cleanup, reinstall, verify claw3d search.

Full guides:

• docs/OPENCLAW_DOCKER.md — General Docker setup
• docs/INSTALL_LOCAL.md — Step-by-step install from local repo (pre-ClawHub)

Troubleshooting: If the bot says it can't convert images to 3D, see docs/TROUBLESHOOTING.md.

Manual (no Docker): Sync skills with ./scripts/sync-skills.sh

ClawHub — When published, use OpenClaw’s skill install (one-click via uv tool install claw3d).

Environment Variables

Variable	Required	Default	Description
FAL_API_KEY	Yes	—	FAL.ai API key
CLAW3D_SLICER_URL	No	http://localhost:8000	CuraEngine API URL
CLAW3D_PRINTER	No	—	Default printer ip:port

Project Structure

claw3d/                    # This repo (claw3d CLI)
├── src/claw3d/            # convert, scale, slice, print, config
├── scripts/
│   ├── sync-skills.sh     # Sync 3d-ai-forger + 3d-printing to workspace
│   └── setup-openclaw-docker.sh
├── reference/             # Architecture notes
├── pyproject.toml
└── README.md
curaengine-slicer-api/     # CuraEngine HTTP API (port 8000, sibling folder)

Related projects

Repo	Description
clarvis-ai	One-command distro — bundles everything into a single Docker Compose stack
claw3d-skill	AI agent instructions that teach OpenClaw how to orchestrate the 3D printing workflow
curaengine-slicer-api	CuraEngine REST API for slicing and mesh conversion

License

MIT

---

For a ready-to-use, anything-to-3D-print experience, check out clarv.is