← Back to Skills Marketplace
nitishgargiitd

3d Cog

by CellCog · GitHub ↗ · v1.0.10 · MIT-0
darwinlinuxwindows ✓ Security Clean
1773
Downloads
3
Stars
5
Active Installs
11
Versions
Install in OpenClaw
/install 3d-cog
Description
AI 3D model generation powered by CellCog. Text-to-3D, image-to-3D — production-ready GLB files for games, AR/VR, e-commerce, and 3D printing. Game assets, p...
README (SKILL.md)

3D Cog - Turn Ideas Into 3D Models

3D model generation from text descriptions or reference images.

Most 3D generation tools need a single, perfectly composed reference image. CellCog takes anything — a text description, a rough sketch, a product photo, even a spreadsheet of 50 items — and handles the entire pipeline: reasoning about what you need, generating optimized reference images, and converting them into production-ready GLB files.

How to Use

For your first CellCog task in a session, read the cellcog skill for the full SDK reference — file handling, chat modes, timeouts, and more.

OpenClaw (fire-and-forget):

result = client.create_chat(
    prompt="[your task prompt]",
    notify_session_key="agent:main:main",
    task_label="my-task",
    chat_mode="agent",
)

All agents except OpenClaw (blocks until done):

from cellcog import CellCogClient
client = CellCogClient(agent_provider="openclaw|cursor|claude-code|codex|...")
result = client.create_chat(
    prompt="[your task prompt]",
    task_label="my-task",
    chat_mode="agent",
)
print(result["message"])

What Makes This Different

Any Input → 3D

The power of CellCog isn't image-to-3D — everyone does that. The power is any-to-any.

What You Send What CellCog Does What You Get
Text description Reasons about the object → generates optimized reference image → converts to 3D Production-ready GLB
Rough sketch Enhances into a clean, detailed reference → converts to 3D Production-ready GLB
Product photo Assesses quality, enhances if needed → converts to 3D Production-ready GLB
High-quality concept art Converts directly to 3D Production-ready GLB
List of 10 items Generates 10 reference images → converts all to 3D 10 GLB files

Batch Generation

Need 10 low-poly weapons for your RPG? 20 furniture models for your room designer? 50 product models for your e-commerce catalog?

One prompt. Multiple 3D models. CellCog's agents generate each reference image with the right composition, angle, and detail level — then convert each to a textured 3D model.

prompt = """
Create 3D models (GLB format) for these 5 fantasy weapons:
1. Enchanted longsword with blue crystal blade
2. Dwarven war hammer with rune inscriptions
3. Elven bow with living vine decorations
4. Shadow dagger with smoke effects on the blade
5. Holy mace with golden sunburst head

Low poly (~10,000 polygons each), game-ready, with PBR materials.
"""

What You Can Create

Game Assets

  • Characters: Heroes, NPCs, enemies, bosses
  • Weapons: Swords, bows, staffs, shields, guns
  • Props: Furniture, treasure chests, potions, tools
  • Vehicles: Cars, spaceships, boats, mounts
  • Environment pieces: Trees, rocks, buildings, bridges

Product Visualization

  • E-commerce 3D viewers: Let customers rotate and inspect products
  • Product prototypes: Visualize designs before manufacturing
  • Packaging mockups: 3D packaging for marketing materials

AR/VR Objects

  • AR filters and objects: Place 3D objects in real environments
  • VR environments: Furnish virtual spaces with custom objects
  • Interactive experiences: Objects users can inspect and interact with

3D Printing

  • Figurines and miniatures: Tabletop gaming pieces, collectibles
  • Functional objects: Custom tools, brackets, cases
  • Architectural models: Building miniatures, terrain pieces

Education & Training

  • Anatomical models: Organs, skeletal systems, molecular structures
  • Historical artifacts: Museum-quality digital replicas
  • Engineering models: Mechanical parts, assembly visualizations

Output Format

All 3D models are delivered as GLB files (binary glTF) — the universal web standard for 3D:

  • Supported by Unity, Unreal, Godot, Three.js, Babylon.js
  • Works in web browsers via \x3Cmodel-viewer> or Three.js
  • Compatible with Blender, Maya, 3ds Max for further editing
  • Includes textures and materials in a single file

Chat Mode for 3D

Scenario Recommended Mode
Single 3D object from a clear description or image "agent"
Batch generation (5-20 objects from a list) "agent"
Complex game asset pipeline with style consistency "agent team"

Use "agent" for most 3D work. It handles everything from single objects to batch generation.

Use "agent team" when you need cross-asset consistency — like generating a full set of fantasy weapons that all share the same art style, or building a complete room of furniture that matches a design language.


Example Prompts

Single object from description:

"Create a 3D model of a steampunk pocket watch with exposed brass gears, an etched glass face, and a chain attachment. GLB format, high detail."

From a reference image:

"Convert this product photo into a 3D model for our online store: \x3CSHOW_FILE>/photos/sneaker_product.png\x3C/SHOW_FILE>

Output as GLB, enable PBR materials for realistic rendering."

Batch generation:

"Generate 3D models for these 8 pieces of modern furniture:

  1. Minimalist sofa (3-seater, light gray)
  2. Round coffee table (walnut wood, glass top)
  3. Floor lamp (arc style, brass finish)
  4. Bookshelf (5 tiers, oak wood)
  5. Dining chair (Scandinavian, white)
  6. Side table (concrete, cylindrical)
  7. Desk (standing desk, white with birch legs)
  8. TV console (low profile, dark walnut)

All low-poly (~15,000 polygons), with PBR materials. GLB format."

From a rough sketch:

"Here's my rough sketch of a robot character: \x3CSHOW_FILE>/sketches/robot_concept.jpg\x3C/SHOW_FILE>

Turn this into a polished 3D model. It's a friendly service robot — round body, simple limbs, LED face display. Style: Overwatch/Pixar clean 3D. Output as GLB."

Game asset set:

"I'm building a dungeon crawler. Create 3D models for these dungeon props:

  • Wooden treasure chest (open and closed variants)
  • Iron torch holder with flame
  • Stone altar with carved runes
  • Wooden barrel (intact and broken)
  • Skull pile

Style: Dark fantasy, hand-painted textures. Low poly for mobile game (~8,000 polygons each)."


Tips for Better 3D Models

  1. Be specific about materials: "brushed aluminum", "aged leather", "polished marble" — CellCog uses these to generate better reference images and textures.

  2. Specify your target platform: "low-poly for mobile game" vs "high-detail for cinematic render" changes the approach completely.

  3. Send reference images when possible: Even imperfect references give CellCog a head start over pure text descriptions.

  4. For batch jobs, describe style once: "All in a cohesive hand-painted fantasy style" keeps your assets consistent.

  5. Request PBR materials for realism: If you need metallic, roughness, and normal maps — say so. Essential for game engines and realistic rendering.


If CellCog is not installed

Run /cellcog-setup (or /cellcog:cellcog-setup depending on your tool) to install and authenticate. OpenClaw users: Run clawhub install cellcog instead. Manual setup: pip install -U cellcog and set CELLCOG_API_KEY. See the cellcog skill for SDK reference.

Usage Guidance
This skill appears to do what it says: it calls CellCog's API to generate GLB files and only needs your CELLCOG_API_KEY and Python. Before installing, confirm you trust cellcog.ai (check their privacy/data-retention and billing policies) because any files or images you send will go to their service. Note there is no install script: the skill assumes a 'cellcog' Python package is available — verify which package/version will be used and prefer official packages from PyPI/GitHub. Use a least-privilege API key if possible, don't supply other credentials, and rotate the key if you stop using the skill. If you need stronger guarantees about where your data is stored or how long it's kept, contact CellCog or test with non-sensitive inputs first.
Capability Analysis
Type: OpenClaw Skill Name: 3d-cog Version: 1.0.10 The 3d-cog skill bundle is a documentation-focused package designed to facilitate 3D model generation via the CellCog service. The SKILL.md file provides clear instructions and Python code examples for using the 'cellcog' library, while the _meta.json defines standard metadata and environment requirements (CELLCOG_API_KEY). There is no executable code or evidence of malicious intent, prompt injection, or unauthorized data access within the provided files.
Capability Tags
requires-sensitive-credentials
Capability Assessment
Purpose & Capability
Name/description (text‑to‑3D, image‑to‑3D) aligns with required items: python3 and a single CELLCOG_API_KEY environment variable. Declared dependency on the 'cellcog' SDK is expected for a Python-based client.
Instruction Scope
SKILL.md instructs the agent to call the CellCog client (create_chat) and to accept uploaded reference files via the usual <SHOW_FILE> placeholders. It does not instruct reading arbitrary system files, other credentials, or exfiltrating data to unexpected endpoints.
Install Mechanism
This is an instruction-only skill with no install spec. It declares a dependency ('cellcog') but provides no install instructions — operationally this assumes the runtime already has the Python package available. This is not an obvious security risk but may cause runtime failures or hide exactly which package/version will be used.
Credentials
Only CELLCOG_API_KEY is required, which matches the external API integration. No unrelated secrets, config paths, or multiple credentials are requested.
Persistence & Privilege
Skill does not request always:true or other elevated persistence. It is user-invocable and allows autonomous invocation (the platform default), which is expected for an integration of this type.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install 3d-cog
  3. After installation, invoke the skill by name or use /3d-cog
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.10
- Added explicit declaration of required runtime dependencies for the skill: Python 3 executable and the CELLCOG_API_KEY environment variable. - These requirements are reflected in the OpenClaw metadata section to ensure proper installation and configuration across supported operating systems. - No other content changes or feature updates.
v1.0.9
- Updated SKILL.md for improved clarity and brevity in installation instructions. - Clarified agent usage examples, changing "Cursor / Claude Code / Other agents" to "All agents except OpenClaw." - Revised installation guidance section to include command variants based on the tool in use. - No changes to core functionality or code; documentation only.
v1.0.8
- Improved and clarified the skill description to highlight AI-powered 3D model generation from text, images, or sketches. - Updated usage instructions with improved code samples, including more explicit import and initialization guidance for various agents. - Enhanced the description of capabilities (text-to-3D, image-to-3D, batch generation) and areas of application. - No changes to core functionality or API; documentation and onboarding are now clearer and more informative.
v1.0.7
- Simplified and shortened the skill description for clarity and focus. - Consolidated and clarified setup instructions, specifying usage in different agent environments. - Added a new section explaining what to do if CellCog is not installed, with setup steps for Cursor, OpenClaw, and other agents. - Removed some repeated intro and guidance, making the readme more concise while retaining all example prompts and key feature explanations. - General content tightening for readability and easier onboarding.
v1.0.6
- Expanded the documentation to cover new use cases, including batch generation, education/training, and asset pipelines. - Clarified the advantages of CellCog’s "any-input to 3D" workflow, emphasizing support for text, sketches, product photos, and batch prompts. - Added usage tips and detailed example prompts for a wider range of 3D asset needs. - Improved explanations for recommended chat modes and how to maximize model quality for various applications. - Updated output format section with compatibility details and editing guidance.
v1.0.5
Version 1.0.5 - Rewrote and condensed SKILL.md for clarity and brevity. - Streamlined feature summaries and removed repetitive marketing/comparison content. - Simplified instructions and descriptions for faster onboarding. - Added a "Related Skills" section for easier skill discovery. - Preserved essential usage details, supported input/output types, and example agent setup.
v1.0.4
- Updated OpenClaw integration instructions for prompt handling and task notification. - Clarified SDK setup requirements and usage for different agent types. - Added examples distinguishing OpenClaw's fire-and-forget mode from blocking modes. - Streamlined references to SDK documentation and API usage.
v1.0.3
- Updated Quick Start instructions in the Prerequisites section for clarity and brevity. - Added a note directing users to the main cellcog skill for full SDK/API documentation, including delivery modes and file handling. - Removed the previous fire-and-forget code snippet and notification explanation to simplify onboarding. - No changes to core capabilities, prompts, or output formats.
v1.0.2
- Updated skill icon emoji from 🧊 to 💎 in metadata. - No functional or documentation content changes.
v1.0.1
- Added supported operating systems (darwin, linux, windows) to metadata. - Added homepage link for CellCog (https://cellcog.ai). - Improved skill description for clarity and broader coverage. - No changes to core features or usage; documentation updated for completeness.
v1.0.0
- Initial release of 3d-cog: Turn sketches, text, or product photos into production-ready 3D GLB models. - Supports input via text descriptions, rough sketches, product photos, or item lists. - Enables batch generation of multiple 3D models from a single prompt. - Integrates with the cellcog skill for streamlined SDK setup and API calls. - Outputs GLB files with textures and materials, compatible with major 3D software and engines. - Includes detailed examples, prompt patterns, and tips to optimize your 3D model requests.
Metadata
Slug 3d-cog
Version 1.0.10
License MIT-0
All-time Installs 5
Active Installs 5
Total Versions 11
Frequently Asked Questions

What is 3d Cog?

AI 3D model generation powered by CellCog. Text-to-3D, image-to-3D — production-ready GLB files for games, AR/VR, e-commerce, and 3D printing. Game assets, p... It is an AI Agent Skill for Claude Code / OpenClaw, with 1773 downloads so far.

How do I install 3d Cog?

Run "/install 3d-cog" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is 3d Cog free?

Yes, 3d Cog is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does 3d Cog support?

3d Cog is cross-platform and runs anywhere OpenClaw / Claude Code is available (darwin, linux, windows).

Who created 3d Cog?

It is built and maintained by CellCog (@nitishgargiitd); the current version is v1.0.10.

💬 Comments