MCPInstall

blender-modeling

by roble3Media & Creative

Create and edit 3D meshes in Blender — primitives, hard-surface modeling, mesh operators, modifier stacks (Bevel, Subdivision, Boolean, Mirror, Array, Solidify), bmesh-level edits, retopology basics. Use whenever the user asks to "make/model/create/build a 3D object", "shape/sculpt this", "add a cube/sphere/cylinder/etc.", "extrude/inset/bevel this face", "add a modifier", or any geometry-creation request that isn't a wireframe trace. Make sure to use this skill even if the user does not say "model" — also covers "make a sword", "build a chair", "add a door", "carve out a hole". Pairs with blender-materials for look-dev and blender-pro-workflow for full pipelines.

Installs
2.3K
Last 24h
+60
Trending rank
#2956
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/roble3/cc-blender-skill --skill blender-modeling
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads blender-modeling automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

MCP backend

Use when: Any geometry creation, mesh edit, or modifier stack work in Blender. Not for wireframe → 3D conversion (use wireframe-to-3d) and not for sculpting strokes (Blender's sculpt mode is gestural, not text-driven).

Before the first Blender call, read ../text-to-blender/references/mcp-backends.md. Prefer official MCP for shared capabilities; preserve the selected backend for the whole task. Tool names in recipes are client/legacy examples: map them through that reference and its agent-host guide.

Blender Modeling

Create geometry in Blender via natural language. You emit Python code that the Blender MCP executes; the patterns below cover the common 80%.

Decision tree

What kind of geometry?
├── Hard-surface (vehicles, weapons, architecture, props)
│   → Cube primitive + Bevel + SubSurf modifier stack
│   → See "Hard-surface stack" recipe
│
├── Organic (characters, creatures, plants — block-out only)
│   → Ico Sphere + sculpting (or Voxel Remesh for shape)
│   → For sculpting strokes, redirect: it's gestural, not text-driven
│
├── Architectural / repeating (fences, columns, tile)
│   → Plane/Cube + Array modifier (+ Curve modifier for paths)
│   → See "Array along curve" recipe
│
├── Cylindrical (pipes, columns, bottles)
│   → Cylinder primitive, or Curve + bevel_object
│   → See "Sweep along path" — covered in wireframe-to-3d if needed
│
├── Holes / cuts in existing geometry
│   → Boolean modifier (DIFFERENCE)
│   → See "Boolean cut" recipe
│
└── Quick block-out from primitives only
    → Multiple primitive_*_add calls
    → See "Block-out scene" recipe

Code-execution rules (recap)

  • Each backend code-execution call gets a fresh Python namespace. Re-import everything; identify objects by bpy.data.objects['name'].
  • Always name objects with GEO- prefix. Never leave Cube.027.
  • Print structured output back so you can parse results.
  • Chunk long sequences into multiple calls.

Recipes

Critical: axis orientation for elongated objects

For any elongated/asymmetric subject (sword blade, knife, bottle, plank, bone, screwdriver tip, etc.), three axes have different meaning:

  • Long axis — the length of the object (78cm for a sword blade)
  • Broad axis — the wider face axis, what's visible from the "useful" viewing angle (4.5cm for a blade — the flat side you'd lay on a table)
  • Thin axis — the narrower cross-section axis (0.8cm for a blade — the cutting edge)

Always orient elongated objects so the broad axis faces the camera in hero shots. A sword viewed edge-on (camera looking down the thin axis) renders as a thin pole and looks nothing like a sword. The recipes below use this convention:

ConventionX (left-right of object's local space)Y (front-back of object's local space)Z (up-down)
Sword bladethin (0.8cm)broad (4.5cm)long (78cm) — vertical
Knife bladethinbroadlong — horizontal
Plankthinbroadlong
Bottlesymmetric (radius)symmetric (radius)long (height)

After building, rotate the object so the broad axis points roughly toward the camera. For a sword standing upright with camera in front (camera in -Y direction): rotate the blade 90° around Z so its local Y (broad) → world X, then the broad face is visible from the camera's perspective.

Critical: connecting parts smoothly (no visible seams)

When assembling a multi-part subject (sword = blade + guard + grip + pommel; chair = seat + back + 4 legs), separate primitives abutting at exactly-aligned face boundaries leave visible seams even though the math says they touch. Worse — different shape primitives (cylinder grip into cube guard) produce obvious "cylinder-on-rectangle" boundaries.

Two fixes, used together:

1. Overlap parts deeply at joins. Make adjacent primitives interpenetrate by 5–15mm at every connection. The hidden volume disappears inside the larger part, leaving no visible seam.

# Sword example: grip extends 1.5cm INTO the guard above and 1cm INTO the pommel below
GRIP_OVERLAP_INTO_GUARD = 0.015
GRIP_OVERLAP_INTO_POMMEL = 0.010
grip_total_len = GRIP_VISIBLE_LEN + GRIP_OVERLAP_INTO_GUARD + GRIP_OVERLAP_INTO_POMMEL

The cylinder grip's top 1.5cm is inside the guard cube — not visible from outside, so the transition you see is just gold-guard surface, no cylinder-meeting-rectangle artifact.

2. Apply shade_smooth() to rounded parts (cylinders, spheres, organic shapes). Shaded-flat cylinders show every facet boundary; smooth-shaded ones look continuous. Cubes and beveled hard-surface parts can stay shaded flat (or be partially smoothed via Auto Smooth on Blender 4.x; Blender 5.x removed Mesh.use_auto_smooth so use modifier-based smoothing or per-face flags).

# After creating each rounded primitive
bpy.ops.object.shade_smooth()

Anti-pattern (visible seams):

# ❌ Pieces abut exactly — visible seam where surfaces meet
pommel_z = -GRIP_LEN/2 - POMMEL_R     # pommel top exactly at grip bottom
guard_z = GRIP_LEN/2 + GUARD_H/2      # guard bottom exactly at grip top
# Result: clear line where each pair of surfaces meets

Correct (hidden seams via overlap):

# ✓ Pieces overlap by ~5-15mm; junction lines are inside other geometry
pommel_z = -GRIP_LEN/2 - POMMEL_R + 0.010   # pommel pushed up 1cm into grip
guard_z = GRIP_LEN/2 + GUARD_H/2 - 0.015    # guard pushed down to envelope grip top

For a truly seamless join (high-quality renders), Boolean Union the same-material parts: e.g. Boolean Union pommel + grip into a single mesh would eliminate the seam entirely. But this only works when both parts use the same material.

Critical: tapering to a point (for blade tips)

Don't just scale the top vertices toward zero — that produces a "chiseled flat" tip. Pinch all top vertices to a single point and merge them:

import bpy
import bmesh

obj = bpy.data.objects['GEO-blade']
bpy.context.view_layer.objects.active = obj
bpy.ops.object.mode_set(mode='EDIT')

bm = bmesh.from_edit_mesh(obj.data)
bm.verts.ensure_lookup_table()

# Find vertices at the top (highest local Z)
max_z = max(v.co.z for v in bm.verts)
top_verts = [v for v in bm.verts if abs(v.co.z - max_z) < 0.001]

# Collapse them to centerline
for v in top_verts:
    v.co.x = 0.0
    v.co.y = 0.0

bmesh.update_edit_mesh(obj.data)

# Merge the now-coincident vertices into a true single point
bpy.ops.mesh.select_all(action='DESELECT')
for v in top_verts:
    v.select = True
bmesh.update_edit_mesh(obj.data)
bpy.ops.mesh.remove_doubles(threshold=0.001)
bpy.ops.object.mode_set(mode='OBJECT')

print(f"tapered:{obj.name}")

This produces a true geometric point. Without remove_doubles, the four collapsed verts stay as four distinct points at the same coordinate — the tip looks visually pointed but is degenerate topology.

Recipe 1 — Add a primitive with a clean name

import bpy

# Add cube
bpy.ops.mesh.primitive_cube_add(size=2.0, location=(0, 0, 1))
obj = bpy.context.active_object
obj.name = 'GEO-base_box'
print(f"created:{obj.name} verts:{len(obj.data.vertices)}")

Replace primitive_cube_add with: _plane_, _uv_sphere_, _ico_sphere_, _cylinder_, _cone_, _torus_, _monkey_. Each takes appropriate arguments (radius, depth, vertices, segments, subdivisions).

Recipe 2 — Hard-surface stack (the "Bevel + SubSurf" pattern)

import bpy

obj = bpy.data.objects['GEO-base_box']

# 1. Bevel modifier — round the sharp edges
bevel = obj.modifiers.new('Bevel', type='BEVEL')
bevel.width = 0.02              # 2 cm round-over
bevel.segments = 3              # smoothness
bevel.limit_method = 'ANGLE'    # only bevel edges sharper than threshold
bevel.angle_limit = 0.523599    # 30° in radians

# 2. Subdivision Surface AFTER bevel (critical order)
subsurf = obj.modifiers.new('SubSurf', type='SUBSURF')
subsurf.levels = 2
subsurf.render_levels = 3

# 3. Smooth shading
bpy.context.view_layer.objects.active = obj
bpy.ops.object.shade_smooth()
print(f"hardsurface:{obj.name}")

Critical: Bevel before SubSurf. Reverse this and you get pinching artifacts.

Recipe 3 — Edit-mode operations (extrude, inset, loop cut)

import bpy

obj = bpy.data.objects['GEO-base_box']
bpy.context.view_layer.objects.active = obj
bpy.ops.object.mode_set(mode='EDIT')

# Select all faces, then extrude up by 1m
bpy.ops.mesh.select_all(action='SELECT')
bpy.ops.mesh.extrude_region_move(
    TRANSFORM_OT_translate={'value': (0, 0, 1.0)}
)

# Inset all selected faces by 0.1m
bpy.ops.mesh.inset(thickness=0.1, depth=0)

# Add a loop cut around the middle
bpy.ops.mesh.loopcut_slide(
    MESH_OT_loopcut={'number_cuts': 1, 'edge_index': 0},
    TRANSFORM_OT_edge_slide={'value': 0.0},
)

bpy.ops.object.mode_set(mode='OBJECT')
print(f"edited:{obj.name} verts:{len(obj.data.vertices)}")

Recipe 4 — Boolean cut (drilling a hole)

import bpy

target = bpy.data.objects['GEO-base_box']
cutter = bpy.data.objects.get('GEO-cutter')

if cutter is None:
    bpy.ops.mesh.primitive_cylinder_add(radius=0.3, depth=3.0, location=(0, 0, 1))
    cutter = bpy.context.active_object
    cutter.name = 'GEO-cutter'

# Apply boolean
mod = target.modifiers.new('Boolean', type='BOOLEAN')
mod.operation = 'DIFFERENCE'
mod.object = cutter
mod.solver = 'EXACT'

bpy.context.view_layer.objects.active = target
bpy.ops.object.modifier_apply(modifier=mod.name)

# Hide cutter from render
cutter.hide_viewport = True
cutter.hide_render = True
print(f"booleaned:{target.name}")

Recipe 5 — Mirror modifier (only model half)

import bpy

obj = bpy.data.objects['GEO-character_half']
mod = obj.modifiers.new('Mirror', type='MIRROR')
mod.use_axis[0] = True   # mirror across X
mod.use_clip = True       # snap vertices on axis
mod.use_mirror_merge = True
mod.merge_threshold = 0.001
print(f"mirrored:{obj.name}")

Place Mirror first in the stack (before Bevel/SubSurf).

Recipe 6 — Array along curve (chains, fences, beads)

import bpy

# 1. The base unit
bpy.ops.mesh.primitive_cube_add(size=0.2, location=(0, 0, 0))
unit = bpy.context.active_object
unit.name = 'GEO-bead'

# 2. The path (assume it exists; user provides or we add a Bezier)
path = bpy.data.objects.get('GEO-path')
if path is None:
    bpy.ops.curve.primitive_bezier_curve_add()
    path = bpy.context.active_object
    path.name = 'GEO-path'

# 3. Array modifier (count or fit to length)
arr = unit.modifiers.new('Array', type='ARRAY')
arr.fit_type = 'FIT_CURVE'
arr.curve = path
arr.relative_offset_displace = (1.0, 0, 0)

# 4. Curve modifier — bends the array along the path
crv = unit.modifiers.new('Curve', type='CURVE')
crv.object = path
crv.deform_axis = 'POS_X'
print(f"arrayed:{unit.name}")

Recipe 7 — Block-out (rapid composition test)

import bpy

# Floor
bpy.ops.mesh.primitive_plane_add(size=10)
bpy.context.active_object.name = 'GEO-floor'

# Hero subject
bpy.ops.mesh.primitive_cube_add(size=1.5, location=(0, 0, 0.75))
bpy.context.active_object.name = 'GEO-subject'

# Background prop
bpy.ops.mesh.primitive_cylinder_add(radius=0.5, depth=2, location=(2, 1.5, 1))
bpy.context.active_object.name = 'GEO-prop_pillar'

print('blockout:done')

Recipe 8 — Cleanup after curve→mesh or boolean

import bpy

obj = bpy.data.objects['GEO-target']
bpy.context.view_layer.objects.active = obj

bpy.ops.object.mode_set(mode='EDIT')
bpy.ops.mesh.select_all(action='SELECT')
bpy.ops.mesh.remove_doubles(threshold=0.0001)
bpy.ops.mesh.normals_make_consistent(inside=False)
bpy.ops.object.mode_set(mode='OBJECT')

bpy.ops.object.shade_smooth()
print(f"cleanup:{obj.name} verts:{len(obj.data.vertices)}")

Modifier stack order (memorize this)

Mirror → Array → Solidify → Bevel → Subdivision Surface → (Boolean if needed)

Wrong order = artifacts. The single most common amateur mistake is SubSurf before Bevel.

Common pitfalls

SymptomFix
Default-cube lookAdd Bevel (0.02m, 3 segments) and SubSurf
Sharp pinch on round shapesBevel before SubSurf, not after
Black faces in renderRecompute normals (mesh.normals_make_consistent)
Boolean creates n-gonsApply Bool, switch to Edit, fix to quads, then SubSurf
Symmetry breaksUse Mirror modifier, not duplicate-and-flip
Mesh has hidden interior facesMesh → Clean Up → Delete Loose

When to load references/overview.md

Load when:

  • The recipes here don't match the request (need bmesh-level precision, custom ops)
  • Topology requirements are stricter than usual (animation-ready, game LODs)
  • Performance matters (foreach_set, batch ops needed)
  • The user references operators not in the recipes

The reference covers: bmesh.ops cookbook, all bpy.ops.mesh.* operators worth knowing, hard-surface workflow with MESHmachine-style chamfering, retopology guidelines, mesh-clean checklist.

What this skill is NOT for

  • Wireframe drawing → 3D model: use wireframe-to-3d
  • Sculpting strokes: Blender's sculpt mode is gestural; can't be driven well from text
  • Sweep-along-path / lofting / curve-driven shapes: covered in wireframe-to-3d/references/blender-patterns.md
  • Materials / lighting / rendering: redirect to those skills

FAQ

What does the blender-modeling skill do?

Create and edit 3D meshes in Blender — primitives, hard-surface modeling, mesh operators, modifier stacks (Bevel, Subdivision, Boolean, Mirror, Array, Solidify), bmesh-level edits, retopology basics. Use whenever the user asks to "make/model/create/build a 3D object", "shape/sculpt this", "add a cube/sphere/cylinder/etc.", "extrude/inset/bevel this face", "add a modifier", or any geometry-creation request that isn't a wireframe trace. Make sure to use this skill even if the user does not say "model" — also covers "make a sword", "build a chair", "add a door", "carve out a hole". Pairs with blender-materials for look-dev and blender-pro-workflow for full pipelines.

How do I install blender-modeling?

Run npx skills add https://github.com/roble3/cc-blender-skill --skill blender-modeling in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/blender-modeling/ (personal) or .claude/skills/blender-modeling/ (project).

Which AI agents can use blender-modeling?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is blender-modeling safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains blender-modeling?

It is published by roble3 in roble3/cc-blender-skill. It has 2,311 recorded installs.

More from roble3/cc-blender-skill

View all →

blender-materials

roble3/cc-blender-skill

Create and assign PBR materials in Blender via Principled BSDF — metals, glass, plastic, fabric, skin, organics. Covers physically-based material recipes with real-world values, Coat layer (varnish/car paint), Sheen (cloth), Subsurface scattering (skin/wax), Transmission (glass), and procedural patterns (wood grain, marble, fabric weave). Use whenever the user asks to "make it look like X material", "give it a metallic finish", "apply a wood texture", "make this glass / plastic / brushed steel / leather / skin", or any look-development request. Make sure to use this skill even if the user does not say "material" — also covers "make it shiny", "matte finish", "looks like copper", "rough surface". Works with any geometry; pairs with blender-lighting (materials only look right under proper lighting).

1.7K installs

blender-export

roble3/cc-blender-skill

Export Blender scenes to glTF/GLB (web/AR), FBX (game engines), OBJ (universal), USD (VFX pipelines), STL (3D printing). Includes Unreal collision hulls and Multires normal-map baking for low-poly export, per-format settings, embed/unpack textures, axis conversion, polygon optimization (Decimate), and target-platform validation. Use whenever the user asks to "export this", "save as glTF / FBX / OBJ / STL / USD", "package for Unity / Unreal / Three.js / web / AR / 3D print", or any output format conversion. Make sure to use this skill even if the user does not say "export" — also covers "package this for the web", "make it work in Unity", "send to Unreal", "save for 3D printing".

Media & Creative1.4K installs

blender-rendering

roble3/cc-blender-skill

Render Blender scenes with the right engine and settings — Cycles for photoreal, EEVEE for speed/stylized, sample counts, denoising (OptiX/OIDN), light path tuning, color management (AgX/Filmic), file output (PNG/EXR/MP4), animation rendering. Use whenever the user asks to "render this", "produce an image", "render a frame / animation", "make a final image", "save the render", or any output-generation request. Make sure to use this skill even if the user does not say "render" — also covers "make a picture", "save the result", "produce a final image", "export as image".

Media & Creative1.4K installs

blender-animation

roble3/cc-blender-skill

Animate objects, cameras, lights, and properties in Blender — keyframes, F-curves, easing (Bezier, Linear, Sine, Bounce, Elastic), shape keys (morph targets / blendshapes / visemes), drivers (Python expressions on properties), NLA actions for reuse and layering. Use whenever the user asks to "animate this", "make it move / rotate / scale over time", "add keyframes", "loop / oscillate", "shape key / morph / blendshape", "visemes for lip sync", or any time-based property change. Make sure to use this skill even if the user does not say "animate" — also covers "spin it slowly", "make it wave", "fade in / out", "pulse", "facial expression".

Frontend & Design1.3K installs

blender-pro-workflow

roble3/cc-blender-skill

End-to-end production workflow guidance for Blender — the order to assemble scenes (block-out → camera → light → forms → materials → detail → render → composite → export), critique protocols, time budgets, and recovery patterns. Use whenever the user asks to "make a complete scene / hero shot / production-quality render", "what's the right order to do this", "set up a full pipeline", or has a multi-step request crossing modeling + lighting + materials + rendering. Make sure to use this skill for any request that spans multiple phases of 3D work, even if the user does not say "workflow" — also covers "make a final image of X", "produce a hero render", "professional-looking result".

Productivity & Automation1.3K installs

reference-to-3d

roble3/cc-blender-skill

Reconstruct Blender models from supplied reference sheets, branding templates, texture atlases, orthographic front/side/back/top views, or mascot/logo art where visual fidelity to the source is more important than a plausible generated object. Use when the user says the model must match a template, wireframe, texture pack, character sheet, mascot sheet, or brand asset exactly; also use after feedback like "does not look like the reference", "fit the texture 1:1", "wrong number of visible parts", or "compare against the template". Requires Blender MCP plus local Python with Pillow/OpenCV/numpy; pairs with blender-uv-texturing, wireframe-to-3d, blender-modeling, blender-materials, and blender-export.

Media & Creative1.3K installs

Popular media & creative skills

View category →

hyperframes

heygen-com/hyperframes

Read first to make, edit, animate or render any video with HyperFrames (HTML rendered to video), including launch films, explainers, music videos, charts, maps, vertical clips, title cards, captions and overlays. Holds the render contract, the work loop and what each kind of video needs.

Media & Creative1.2M installs

media-use

heygen-com/hyperframes

Get or make any media for a HyperFrames video - music, sound effects, images, icons, logos, voiceover, avatar and generated video, transcripts, cutouts, colour grades and looks - with HeyGen or free local tools, and mix placed audio. Also for footage that looks dark, flat or dull, or should be stylized or hidden.

Media & Creative1.1M installs

hyperframes-registry

heygen-com/hyperframes

About 450 ready-made HyperFrames blocks and components installed as editable source - transitions, overlays, titles, kinetic type, captions, charts, map overlays, 3D, camera moves. Use it before hand-building any named visual (CRT scanlines, glitch, film grain, a chart); also for hyperframes add and catalog.

Media & Creative1M installs

hyperframes-creative

heygen-com/hyperframes

Non-animation creative direction for HyperFrames videos. Use for design spec (frame.md / design.md) handling, palettes, typography, narration, beat planning, audio-reactive visuals, composition patterns, and brand / style decisions. For atomic motion patterns and scene blueprints, use hyperframes-animation.

Media & Creative916.5K installs

ai-video-generation

101-skills/superpowers

Generate AI videos with Google Veo, Seedance 2.0, HappyHorse, Wan, Grok and 40+ models via inference.sh CLI. Models: Veo 3.1, Veo 3, Seedance 2.0, HappyHorse 1.0, Wan 2.5, Grok Imagine Video, OmniHuman, Fabric, HunyuanVideo. Capabilities: text-to-video, image-to-video, reference-to-video, video editing, lipsync, avatar animation, video upscaling, foley sound. Use for: social media videos, marketing content, explainer videos, product demos, AI avatars. Triggers: video generation, ai video, text to video, image to video, veo, animate image, video from image, ai animation, video generator, generate video, t2v, i2v, ai video maker, create video with ai, runway alternative, pika alternative, sora alternative, kling alternative, seedance, happyhorse

Media & Creative881.5K installs

ai-image-generation

101-skills/superpowers

Generate AI images with GPT-Image-2, FLUX, Gemini, Grok, Seedream, Reve and 50+ models via inference.sh CLI. Models: GPT-Image-2, FLUX Dev LoRA, FLUX.2 Klein LoRA, Gemini 3 Pro Image, Grok Imagine, Seedream 4.5, Reve, ImagineArt. Capabilities: text-to-image, image-to-image, inpainting, LoRA, image editing, upscaling, text rendering. Use for: AI art, product mockups, concept art, social media graphics, marketing visuals, illustrations. Triggers: flux, image generation, ai image, text to image, stable diffusion, generate image, ai art, midjourney alternative, dall-e alternative, text2img, t2i, image generator, ai picture, create image with ai, generative ai, ai illustration, grok image, gemini image, gpt image, openai image, chatgpt image

Media & Creative880.3K installs