blender-design
Skill4 starsChanged 21 days agoApache-2.0View on GitHub
Turn a user's idea into a Blender scene through milestone-based modeling, materials, lighting, camera, and animation commands.
In this repository44 other files it gives its agents.
- AGENTS.mdAGENTS.md
- openspec-apply-changeSkill
- openspec-archive-changeSkill
- openspec-exploreSkill
- openspec-proposeSkill
- openspec-sync-specsSkill
- openspec-update-changeSkill
- blender-ai-replicationSkill
- blender-asset-librarySkill
- blender-background-jobsSkill
- blender-character-animationSkill
- blender-character-riggingSkill
- blender-cinematographySkill
- blender-connectorSkill
- blender-curvesSkill
- blender-design-loopSkill
- blender-exportSkill
- blender-grease-pencilSkill
- blender-hairSkill
- blender-hard-surfaceSkill
- blender-harness-drivingSkill
- blender-harnessSkill
- blender-inspectSkill
- blender-managedSkill
- blender-mcp-setupSkill
- blender-previewSkill
- blender-previsSkill
- blender-procedural-modelingSkill
- blender-quality-validationSkill
- blender-recoverSkill
- blender-render-compositingSkill
- blender-retopologySkill
- blender-scene-assemblySkill
- blender-sculpt-surfaceSkill
- blender-seedance-pipelineSkill
- blender-sequence-editingSkill
- blender-simulationSkill
- blender-to-dreaminaSkill
- blender-trackingSkill
- blender-useSkill
- blender-uv-materialSkill
- blender-video-originalSkill
- blender-video-recreateSkill
- blender-visual-loopSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Prefer registered commands over expert Python.
- Distinctive: line 31
Default cadence is plan-first autonomy.
- Very specific: line 44
Read foreground control and recovery.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Blender Design Workflow
Inspect, then create an implementation brief before mutation: requested assets, supplied references, missing references, scene constraints, camera route, animation beats, duration, and required exports. If a required reference is missing, ask the user for it by default. Create a Blender-designed proxy only when the user explicitly requests design of the missing asset, and record its assumptions and deviations in the delivery receipt.
For an auto_with_budget request, execute the approved brief through every safe milestone
without asking after each preview. Capture the same previews and validation evidence, but present
them together with the final artifact inventory. Stop only for a forbidden missing asset, path
escape, deletion/overwrite, expert Python or failed validation requiring recovery. Remote budget
enforcement is owned by the downstream plugin.
Translate the approved brief into named components, then complete Scene Structure, Modeling,
Materials, Lighting and Camera, Animation, and Final Preview milestones. Use one transaction per
milestone and expectedSceneRevision on every mutation.
Delegate each milestone to the precise domain Skill advertised by capability.describe; do not
keep advanced animation, cinematography, quality validation, background jobs, or simulation under
this general workflow. Multi-domain recipes may load multiple Skills declared by the capability.
Prefer registered commands over expert Python. Read session.status and show the actual scene
in the foreground before work. Update session.set_progress with an honest stage and optional
progress fraction; report actual objects changed, not a synthetic completion percentage.
Default cadence is plan-first autonomy. Before mutating the scene, present an overall plan
(structure, milestone order, deliverable formats, output location) and get ONE confirmation;
unless the user opts into interactive review, then run milestones end to end in
auto_with_budget mode: after each milestone generate fresh previews, assess them against the
approved brief, and commit without another prompt. Only irreversible operations (delete,
overwrite, expert Python, path escape, budget exceeded) interrupt for approval.
On failure recover the transaction only if the user has not taken over. Command success alone is not
design acceptance. Before final export, verify all brief constraints that are observable in the
scene and preview: object count and uniqueness, animation beat order, frame range, camera route,
and required deliverable formats. Return the artifact inventory and explicitly ask whether the
user wants to finish locally or hand off only when that choice was not already made.
Downstream AI rendering is outside this Skill.
Read foreground control and recovery.
On SESSION_PAUSED, stop design commands. Only a user resume permits continuation; then inspect
again and begin a new transaction. Never roll back the pre-takeover transaction over user edits.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

