---
type: "post"
id: "blender-skills-ships-22-agent-skills-covering-blender-03950e"
slug: "blender-skills-ships-22-agent-skills-covering-blender-03950e"
title: "blender-skills ships 22 Agent Skills covering Blender 5.2 LTS modeling, rigging, baking, rendering, export and QA."
description: "The blender-skills repository packages 28 Blender workflow skills for agents including Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor."
retrieval_nugget: "The blender-skills repository packages 28 Blender workflow skills for agents including Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor. Its instructions span much of a 3D asset pipeline, from modeling to quality checks. For teams using agents on Blender work, it offers a ready-made starting point to test and adapt."
published_at: "2026-11-15"
updated_at: "2026-11-15"
record_date: "2026-11-15"
date_kind: "published_at"
topics: ["claude","claude-code","codex"]
entities: ["EL4CTEO","blender-skills","Blender","Claude Code","Codex","OpenCode","Gemini CLI","Cursor"]
related_patterns: ["skills-become-portable-capability-layer"]
novelty_label: "new_fact"
source_url: "https://github.com/EL4CTEO/blender-skills"
source_urls: ["https://github.com/EL4CTEO/blender-skills"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:blender-skills-ships-22-agent-skills-covering-blender-03950e"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/blender-skills-ships-22-agent-skills-covering-blender-03950e/","markdown":"https://newruntime.com/posts/blender-skills-ships-22-agent-skills-covering-blender-03950e.md","json":"https://newruntime.com/posts/blender-skills-ships-22-agent-skills-covering-blender-03950e.json"}
markdown_url: "https://newruntime.com/posts/blender-skills-ships-22-agent-skills-covering-blender-03950e.md"
---

# blender-skills ships 22 Agent Skills covering Blender 5.2 LTS modeling, rigging, baking, rendering, export and QA.

## Retrieval answer

The blender-skills repository packages 28 Blender workflow skills for agents including Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor. Its instructions span much of a 3D asset pipeline, from modeling to quality checks. For teams using agents on Blender work, it offers a ready-made starting point to test and adapt.

## Summary
The blender-skills repository packages 28 Blender workflow skills for agents including Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor. Its instructions span much of a 3D asset pipeline, from modeling to quality checks. For teams using agents on Blender work, it offers a ready-made starting point to test and adapt.

## What happened
The EL4CTEO/blender-skills GitHub repository describes itself as a collection of professional Blender workflows packaged as Agent Skills. Its README lists 28 skills covering modeling, UV work, texturing, baking, rigging, animation, clothing, hair, environments, optimization, export, rendering, video, simulation and QA. The project says it is built for Blender 5.2 LTS and newer, and that every skill is verified on Blender 5.3. It recommends updating to Blender 5.3.0 for the best experience, citing KTX2 web export, a compositor effects stack and Gaussian splats as additions in that release.

The skills divide the work into named problem areas. Examples include `blender-core`, which runs code, routes tasks and sets out Blender 5.x API rules; `blender-retopology-uv`, for retopology, UVs and texel density; `blender-texture-baking`, for normal, ambient-occlusion and curvature maps; and `blender-export-pipelines`, for Roblox, Unity, Unreal, Godot, web and USD. The list also includes asset QA checks for issues such as z-fighting, floating parts, bad normals, weights and textures, plus workflow recipes that the repository describes as complete asset playbooks.

The README names Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor as supported agents. Claude Code users can install through a plugin marketplace command. For other agents, the documented route is to clone the repository and run its installer. By default, it installs into `~/.claude/skills` and `~/.agents/skills`; options include selecting agents, installing into a project directory and uninstalling. The repository says `.agents/skills` is read by Codex, OpenCode, Gemini CLI, Copilot and Cursor. Blender MCP is optional: with a Blender MCP server, agents can work in an open Blender session; without one, they run Blender in the background.

For development, the repository documents commands to validate skill specifications and token budgets, and to run tests. It says the test suite includes 67 headless Blender tests, with `BLENDER_EXE` available to override the Blender executable. GitHub displayed an MIT license and 1.2k stars when the page was captured. The README proposes example tasks such as creating a game-ready crate with baked normals and LODs, rigging a character for Unity export, and checking a model before Roblox export.

## What the archive adds
New Runtime's earlier coverage described [74 game-development skills for AI coding agents](https://newruntime.com/posts/awesome-gamedev-agent-skills-74-game-dev-skills-edfd71/) on 2026-10-24 and [Remotion's problem-area skill packages](https://newruntime.com/posts/remotion-skills-2-0-ships-sub-skills-interactivity-and-taste-neutral-routing/) on 2026-09-18. This is a new project and a new domain-specific collection within the broader move to package task guidance as agent skills.

## Trend signal
This supports [skills-become-portable-capability-layer](https://newruntime.com/patterns/skills-become-portable-capability-layer/): the project packages specialized instructions for several agent environments and exposes them through an installer. The practical test is whether builders can inspect, select and maintain the guidance as a reusable capability, rather than treating a large instruction bundle as a substitute for workflow-specific verification.

## Why now
The repository targets Blender 5.2 LTS and newer, and specifically recommends Blender 5.3.0 for the best experience. That makes compatibility with the current Blender feature set part of the project's pitch, rather than presenting the skills as version-agnostic instructions.

## Why it matters
A 3D asset workflow crosses many distinct tasks: geometry, UVs, materials, rigging, animation, optimization and export all have different constraints. A skill collection organized around those tasks can give an agent more relevant guidance than a single generic instruction, while letting a team try it at a particular point in its pipeline. The repository's support for several agents also makes the package worth evaluating independently of one agent vendor.

That is a starting point, not evidence that the workflows reliably produce production-ready assets. Teams still need to check outputs against their own conventions, target engine requirements and quality bar. The repository's QA skill and headless tests are useful signals of an effort to make parts of the workflow testable, but they do not establish the quality of every generated model or the effectiveness of every skill. Start with one repeated task, inspect its results and adapt the instructions before expanding use.

## Limits and uncertainty
The repository describes its skills as verified on Blender 5.3, but that does not independently establish how they perform across different projects, agents or operating environments. The README mentions a comparison of outputs with and without skills, but the page text does not provide its results. It also does not quantify time saved, error rates or the amount of human review required. Although it lists 67 headless tests, their existence alone does not show how much of the full 3D pipeline those tests cover.

## Skill to build now
Pick one repeatable asset task—such as baking normal and ambient-occlusion maps or checking an export—and run it once with the relevant repository skill and once with your current workflow. Keep the same input asset and acceptance criteria. Record what required correction, whether the export passed your checks, and which instructions need project-specific changes before you let the skill handle more of the pipeline.

## Reader takeaway
blender-skills is a multi-agent, MIT-licensed package of task-specific Blender guidance, not a new Blender model or a guarantee of finished assets. Its value for a team depends on how well each skill matches the team's pipeline and how carefully outputs are checked. Test one repeated workflow before adopting the collection broadly.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub.
- topic hub: [Claude Code](https://newruntime.com/topics/claude-code.md): Explore the claude-code topic hub.
- topic hub: [Codex](https://newruntime.com/topics/codex.md): Explore the codex topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)
