Blender Core logo

Blender Core

Community
EL4CTEO
blender-core

Foundation for any Blender 5.2+ task an agent drives - running bpy via MCP (live) or headless CLI, context overrides, locale-safe node lookup, enum discovery, 5.x API changes, verification. Use first for any Blender, bpy, .blend or 3D-asset request; routes to the specialized blender-* skills.

Overview

PublisherEL4CTEO
Repositoryblender-skills
Skill nameblender-core
Stars
426
Forks
39
Bundled files
5
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 5 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by EL4CTEO on GitHub. Read the source before you install it.

Installation

Install the Blender Core AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/EL4CTEO/blender-skills/tree/main/skills/blender-core
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/EL4CTEO/blender-skills.git /tmp/blender-skills
mkdir -p .claude/skills
cp -r /tmp/blender-skills/skills/blender-core .claude/skills/blender-core
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Blender Core in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Blender Core on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Blender Core is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Blender Core (5.2 LTS+)

Route to the specialized skill

TaskSkill
Units, collections, naming, linking, asset librariesblender-scene-setup
Hard-surface modeling, bevels, booleans, normalsblender-hardsurface-modeling
Sculpting, multires, remeshblender-sculpting
Retopology, UV unwrap, texel densityblender-retopology-uv
PBR materials, color spaces, texture setsblender-pbr-materials
Baking high→low maps (normal, AO, curvature, ...)blender-texture-baking
Geometry Nodes, procedural modelingblender-geometry-nodes
Body rigging, Rigify, weight paintingblender-character-rigging
Face rig, shape keys, ARKit, visemesblender-facial-shapekeys
Keyframes, actions, NLA, baking, root motionblender-animation
Poly budgets, LODs, draw calls, cleanupblender-game-optimization
Export: glTF/FBX/USD to Roblox, Unity, Unreal, Godot, webblender-export-pipelines
Cycles/EEVEE settings, compositor, batch renderblender-rendering
Lighting, HDRI, camera, turntableblender-lighting-lookdev
Time of day, night, sunset, fog, mood (cyberpunk, horror...)blender-time-of-day
Rain, snow, wet surfaces, puddles, lightning, dust/embersblender-weather-fx
Cloth, rigid body, fluids, hair, GN physicsblender-simulation
Writing add-ons / extensionsblender-addon-development
Batch/headless pipelines, file conversionblender-pipeline-automation
Checking a mesh/asset before delivery (always)blender-asset-qa
End-to-end playbooks (prop, character, Roblox, product shot)blender-workflow-recipes

Execute

Live (blender-mcp execute_blender_code): work in small steps and print only what you need. To call a skill script, import it:

python
import sys, importlib; sys.path.insert(0, r"<skills_dir>/blender-asset-qa/scripts")
import qa_check; importlib.reload(qa_check); print(qa_check.run(objects=["Chair"]))

Each MCP call runs in a fresh namespace: re-import in every call. That's cheap because modules stay cached in sys.modules. reload only after editing a script. Headless (any agent with a shell): this prints only the script's JSON (@@JSON@@ lines):

python <skills_dir>/blender-core/scripts/run_in_blender.py --blend in.blend <script.py> -- --objects Chair --save out.blend
python <skills_dir>/blender-core/scripts/run_in_blender.py --expr "import bpy; print(len(bpy.data.objects))"

Raw form: blender -b in.blend --factory-startup --python-exit-code 1 -P x.py -- args. Every script here exposes run(**kw) -> dict and the same flags after --. First call in a session: scripts/env_report.py, which returns the version, engines, GPU, add-ons, units, color management and scene counts.

Rules

  • Data API over operators: use bpy.data, bmesh, mesh.attributes, and foreach_get/set with numpy. Operators depend on context and selection, and many need a 3D View, which background mode doesn't have.
  • If an operator is unavoidable: with bpy.context.temp_override(object=ob, active_object=ob, selected_objects=[ob], selected_editable_objects=[ob]): bpy.ops.... Check that the result is {'FINISHED'}.
  • Edit mode isolation: ob.data isn't updated while in Edit Mode. Use bmesh.from_edit_mesh + bmesh.update_edit_mesh, or switch to OBJECT mode before reading.
  • Nodes: find them by type (n.type == "BSDF_PRINCIPLED"), never by name, because names are localized. Sockets by English name or identifier are fine.
  • Enums: read the valid values with [i.identifier for i in T.bl_rna.properties[p].enum_items]. Dynamic enums (render.engine, view_transform, look) under-report, so assign them inside try/except TypeError (the message lists the valid values).
  • Engines: BLENDER_EEVEE, CYCLES, BLENDER_WORKBENCH. Cycles device: scene.cycles.device = "GPU" + preferences compute_device_type.
  • Evaluated data: use ob.evaluated_get(depsgraph), then .to_mesh(), then always .to_mesh_clear(). Call view_layer.update() before reading matrix_world after a change.
  • Freed data: applying or removing a modifier, or removing an object or ID, frees it. Collect names first ([m.name for m in ob.modifiers]) and never touch the old Python reference afterward, because it returns garbage or crashes.
  • Never trust default names ("Cube.001"). Keep object references, rename explicitly, and look objects up via bpy.data.objects[name].
  • Scale: 1 unit = 1 m. Apply rotation/scale (transform_apply) before bevels, modifiers with distances, rigging, physics, baking and export.
  • Paths: bpy.path.abspath("//textures/x.png") resolves relative to the .blend. Save explicitly; use save_as_mainfile(filepath=..., copy=True) for backups.
  • Live sessions are the user's work: never read_factory_settings, never delete data you didn't create, and save a backup copy before destructive batch edits.
  • GPU module in background: call gpu.init() first (5.2+). bgl no longer exists.
  • Big meshes: no per-vertex bpy.ops. Use numpy foreach_get (see references/bpy-patterns.md).

5.x API changes that break old tutorials/LLM memory

action.fcurves → channelbags · scene.node_tree → scene.compositing_node_group · material.use_nodes is deprecated (new materials already have nodes) · mod["Socket_2"] → mod.properties.inputs.Socket_2.value (5.2) · BLENDER_EEVEE_NEXT → BLENDER_EEVEE · Boolean FAST → FLOAT · bone layers → bone collections · bl_info add-ons → extensions (blender_manifest.toml) · bpy.props values are no longer dict-accessible (scene["cycles"]). Full table: references/api-changes-5x.md.

Verify (required before saying "done")

  1. Numbers, not adjectives: report tris, objects, materials, dimensions (ob.dimensions) and the output files written.
  2. Geometry changed → run blender-asset-qa qa_check.py (z-fighting, detached parts, non-manifold, flipped normals, scale).
  3. Look at it: MCP look (mode="angles", shading material/wireframe). Headless: blender-asset-qa/scripts/review_shots.py renders contact sheets.
  4. Re-read your own output: confirm that exported files exist and that their size is non-zero.

References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Blender Core AI skill do?

Foundation for any Blender 5.2+ task an agent drives - running bpy via MCP (live) or headless CLI, context overrides, locale-safe node lookup, enum discovery, 5.x API changes, verification. Use first for any Blender, bpy, .blend or 3D-asset request; routes to the specialized blender-* skills.

Why use Blender Core on TypingMind?

Because you install it once and use it with any model. Blender Core is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Blender Core in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/EL4CTEO/blender-skills/tree/main/skills/blender-core. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Blender Core?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Blender Core?

As many as you like. As long as a model supports skills, you can use Blender Core with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Blender Core AI skill free?

Yes. It is published on GitHub by EL4CTEO under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇