3dprint

name	3dprint
description	Turn natural language into 3D prints on a Bambu Lab printer. ONLY activate when the user EXPLICITLY asks to 3D print, make, or fabricate a physical object. Do NOT activate for general questions, design discussions, or unrelated requests.
homepage	https://github.com/phoenixjyb/openclaw-3dprint
metadata	{"openclaw":{"emoji":"🖨","os":["darwin","linux"],"requires":{"bins":["3dprint"]},"install":[{"id":"pip","kind":"pip","package":"openclaw-3dprint","bins":["3dprint"],"label":"Install openclaw-3dprint (pip)"}]}}

3dprint — Text-to-3D-Print Pipeline

Turn a chat message like "print me a small vase" into a physical 3D print, fully automated.

Quick Start

1. Install: pip install openclaw-3dprint
2. Copy config: cp .env.example ~/.openclaw-3dprint/pipeline.env and fill in your keys
3. Start the pipeline server: openclaw-3dprint --mode feishu
4. Use the CLI from OpenClaw: 3dprint request a small cinderella figurine

CLI Commands

3dprint request <description>   # Submit a new 3D print request
3dprint approve <job_id>        # Approve the current pipeline stage
3dprint reject <job_id>         # Reject/cancel the current stage
3dprint status [job_id]         # Show job status (all or specific)
3dprint queue                   # Show printer queue status
3dprint help                    # Show help

Pipeline Stages

Each print job goes through these stages (approval required between each):

1. LLM Interpretation — enriches the user prompt into a detailed 3D modeling prompt
2. 3D Model Generation — generates a mesh via Tripo3D (or Meshy.ai)
3. Slicing — slices the model with PrusaSlicer (local) or Bambu Studio (remote Windows)
4. Printing — uploads .3mf via FTPS and starts print via MQTT

Configuration

All config is read from ~/.openclaw-3dprint/pipeline.env. Required settings:

Setting	Description
OPENAI_API_KEY	LLM API key (OpenAI, xAI/Grok, etc.)
TRIPO_API_KEY	Tripo3D API key for mesh generation
BAMBU_PRINTER_IP	Printer IP address on local network
BAMBU_PRINTER_SERIAL	Printer serial number
BAMBU_PRINTER_ACCESS_CODE	Printer access code (from LCD screen)

Optional settings:

Setting	Default	Description
BOT_MODE	feishu	Bot mode: feishu, telegram, or dual
SLICER_MODE	local	local (PrusaSlicer) or remote (Windows SSH)
SLICER_PATH	auto-detect	Path to PrusaSlicer binary
BAMBU_SEND_METHOD	ftp	ftp (direct) or studio (Bambu Studio CLI)
MESH_PROVIDER	tripo	tripo or meshy

See .env.example for the full list.

Printer Queue

Multiple agents/users share one printer safely. The pipeline uses cross-process locking (fcntl.flock) so concurrent print requests are serialised. Users see their queue position while waiting.

When to Activate

ONLY trigger this skill when the user's message contains a clear intent to 3D print a physical object. Look for explicit keywords:

• ✅ Trigger: "3D print me a …", "print out a …", "打印一个…", "帮我打一个…", "make me a … on the printer"
• ✅ Trigger: "check print status", "printer queue", "打印状态"
• ❌ Do NOT trigger: "what does a dragon look like", "design a vase" (discussion, not printing)
• ❌ Do NOT trigger: "print this document", "print the report" (paper printing, not 3D)
• ❌ Do NOT trigger: any message without explicit 3D print / 打印 intent

When in doubt, ask the user "Would you like me to 3D print this?" rather than triggering the pipeline.

Typical Agent Interaction

When a user says "3D print me a dragon", you (the agent) should:

1. Enrich the prompt yourself — you already have an LLM. Turn "a dragon" into a detailed 3D modeling description (shape, pose, proportions, style, scale ~120mm, material PLA).
2. Call the API with both fields:

   curl -X POST http://127.0.0.1:8765/api/print \
     -H 'Content-Type: application/json' \
     -d '{"prompt": "a dragon", "enriched_prompt": "A fierce Western dragon in a rearing pose, wings spread wide..."}'
   

   This skips the built-in LLM step — no separate LLM API key needed.
3. If you send only prompt (no enriched_prompt), the pipeline uses its own LLM (requires OPENAI_API_KEY in config).
4. After each stage, show the result and ask user to approve: 3dprint approve <id>
5. When printing starts, report progress via MQTT updates

Requirements

• Python ≥ 3.11
• PrusaSlicer (for local slicing): brew install --cask prusa-slicer
• Network access to Bambu printer (same LAN, ports 990 + 8883)
• API keys: LLM provider + Tripo3D