blender-setup
Skill11 starsChanged 36 days agoMITView on GitHub
Quickly check the official Blender Lab MCP connection using fresh Claude session evidence, deepen failed checks, and report AI agent, Python, Blender and editor readiness without scene changes.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- haw2fregel-lab/claude-in-blendertheir own version, first seen Aug 24, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- haw2fregel-lab/claude-in-blender★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository30 other files it gives its agents.
- AGENTS.mdAGENTS.md
- blender-convert-2d-3dSkill
- blender-convert-3d-2dSkill
- blender-create-environmentSkill
- blender-create-greyboxSkill
- blender-create-modelSkill
- blender-create-visual-target-2dSkill
- blender-game-exportSkill
- blender-light-cameraSkill
- blender-materialsSkill
- blender-procedural-geometrySkill
- blender-renderSkill
- blender-review-optimizeSkill
- blender-rig-animateSkill
- blender-uv-bakeSkill
- blender-convert-2d-3dSkill
- blender-convert-3d-2dSkill
- blender-create-environmentSkill
- blender-create-greyboxSkill
- blender-create-modelSkill
- blender-create-visual-target-2dSkill
- blender-game-exportSkill
- blender-light-cameraSkill
- blender-materialsSkill
- blender-procedural-geometrySkill
- blender-renderSkill
- blender-review-optimizeSkill
- blender-rig-animateSkill
- blender-setupSkill
- blender-uv-bakeSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use /blender-setup to audit the active AI agent's official Blender Lab MCP connection.
- Distinctive: line 21
Check official MCP installation, enablement/compatibility and bridge runtime as three distinct report rows at the effective host/port.
- Distinctive: line 27
Use title case for report step labels, headings and readiness labels.
- Distinctive: line 31
Use exact headers Step, Status, Comment and these rows:
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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 Setup
Use /blender-setup to audit the active AI agent's official Blender Lab MCP connection. Default to a read-only check; repairs, installations, configuration edits and application startup require the user's requested scope.
Use the existing MCP connection whenever it can perform the operation. On Windows, auxiliary execution MUST remain windowless, including brief flashes, during setup, diagnostics, rendering, export, and verification. Before launching, establish suppression for the complete process chain, including client-owned and dependency-owned startup. Use explicit no-console creation (such as subprocess.CREATE_NO_WINDOW) and captured output for reviewed console-only commands. If any required boundary is unknown or unsupported, do not launch: report BLOCKED with the reason and next action, and continue independent safe work. Never retry through a visible terminal, toggle Blender's system console, or weaken suppression. An absolute executable path, cmd /c, Blender --background, or Start-Process -WindowStyle Hidden alone is not proof of no flashes. Keep exit status, bounded timeouts, useful diagnostics without secrets, and cleanup of owned descendants. Do not hide, minimize, restore, focus, or close the existing Blender editor to suppress auxiliary windows. Preserve fresh editor-readiness checks before screenshots.
Workflow
- Identify the platform and active Claude session. Check session availability directly; use client-neutral AI agent labels in the report.
- Inspect the actual native tool catalog and perform one allowed read-only Blender query. Inspect MCP and nested Blender errors; a successful handshake remains separate from scene access.
- Inspect Claude's registration through /mcp or safe read-only claude mcp list / claude mcp get when available. Never substitute Codex TOML/helper evidence for Claude's configuration. Print no secrets.
- Report Python Available and Python Configured separately. Infer runtime readiness only from a fresh working official Python stdio connection, labeling the inference. Diagnose the configured interpreter on failure rather than unrelated terminal Python. Recognize uvx as limited/isolated; do not provision it during an audit.
- Identify Blender only from currently running programs. Never search disk, registry, installation directories or PATH for its executable or use BLENDER_PATH discovery. With no running process, installation is unknown.
- Target the Blender instance reached by MCP, using a read-only process ID query or verified bridge owner. Never substitute another instance's visible window; ask which is intended only when identity remains ambiguous.
- Freshly inspect the target editor's visible/non-minimized state without restoring or focusing it. Maximization is unnecessary. Unavailable inspection is unknown/Blocked and does not erase a working MCP connection.
- Check official MCP installation, enablement/compatibility and bridge runtime as three distinct report rows at the effective host/port. A listener alone does not prove identity; an expanded Preferences entry does not prove enablement.
- Begin each invocation with quick fresh evidence; deepen only failed, conflicting or necessary missing evidence in one bounded second pass. Preserve independent passes, retry once only for a clearly transient scene error, and never reuse saved/cached setup status or start another unreviewed client.
- Display all twelve rows plus separate MCP and editor verdicts, concrete solutions and platform limitations. Do not invent unobserved versions or save reports automatically.
Report
Use title case for report step labels, headings and readiness labels; preserve AI, MCP and Python spelling. Use Pass, Fail and Blocked for displayed statuses, and sentence case for comments and next actions. Preserve exact casing in names, identifiers, paths, versions and quoted UI text.
Official MCP Installed means the official Blender Lab add-on is present. Official MCP Configured means it is enabled and compatible with Blender; verify the expected endpoint separately. Official MCP Running means the official bridge is responding. An installed add-on does not prove enablement, and an enabled add-on does not prove bridge startup. If identity or state is unverified, report Blocked; a socket listener alone does not establish an official running bridge. With direct Preferences evidence, an installed but unchecked add-on passes Installed and fails Configured with Check the box beside MCP; an enabled entry showing Start MCP Server fails Running with that startup action.
Use exact headers Step, Status, Comment and these rows:
| Step | Status | Comment |
| 1. AI Agent Available | observed status | Current Claude session evidence |
| 2. AI Agent Configured | observed status | Claude official MCP registration |
| 3. Python Available | observed status | Registered runtime available; label inference |
| 4. Python Configured | observed status | Libraries in that runtime; label inference or unknown |
| 5. Blender Installed | observed status | Running-program evidence; otherwise unknown |
| 6. Blender Running | observed status | Current process evidence |
| 7. Blender Open | observed status | Target editor visible and not minimized |
| 8. Official MCP Installed | observed status | Official Blender Lab add-on installation evidence |
| 9. Official MCP Configured | observed status | Add-on enabled and compatible with Blender |
| 10. Official MCP Running | observed status | Official bridge responding at the expected endpoint |
| 11. MCP Handshake / Tools | observed status | Current session initialization/tool discovery |
| 12. Live Blender Communication | observed status | Fresh read-only scene query |
Replace placeholders with ✅ Pass, ❌ Fail, ⛔ Blocked. Every failed row includes its solution; blocked rows identify unknown evidence and the next safe step. Label inferred passes. If Blender is absent from running programs, installed is Blocked/unknown, running fails with Open Blender, and editor open is Blocked.
Report MCP Readiness and Editor Readiness separately. A minimized editor requires Restore Blender from the taskbar before capture while live MCP access may still pass. Preserve handshake success when a scene query fails. A working scene query does not prove an unread persisted configuration, exact host interpreter/version, or visible window. Blender's embedded Python is not the host MCP server interpreter.
Troubleshooting and Compatibility
Check required libraries in the interpreter Claude actually launches. If PEP 668/externally managed Python rejects an installation, guide the user to a dedicated venv; do not bypass that protection. An environment can use the same host Python as hobby work and does not require another Blender version. Do not install, switch interpreters or rewrite a working registration as an audit side effect.
Inspect the checkbox beside MCP in supplied Preferences screenshots. Expanded details with an unchecked box mean installed but disabled: Check the box beside MCP, then start its bridge when authorized. An enabled entry showing Start MCP Server means the bridge is stopped. A refused socket alone cannot distinguish these cases. Do not recommend reinstalling a visibly installed add-on.
The local add-on TCP bridge need not serve a browser page at localhost:9876. Test MCP/live scene access instead. The architecture is AI client → official Python server over stdio → add-on inside Blender over local TCP. The AI client launches the host server; a separate LLM client inside Blender is unnecessary. Llama.cpp is optional. Use official setup and official releases for compatibility research; do not replace the official server or automatically upgrade. Report version discrepancies without erasing actual live access.
Windows 10/11 share relevant Win32 inspection APIs; inspect the real desktop only through available supported read-only tools. Sandbox isolation or permission gaps can make visibility unknown. macOS support is limited and untested live in this repository's Windows validation: use available native MCP and running-program evidence, report unsupported editor/listener/interpreter diagnostics as Blocked, and do not grant permissions or invent platform support. The Claude package intentionally does not ship the Codex TOML helper.
Editor Readiness and Screenshots
Before every editor/window/area screenshot, freshly check that the target process has a visible, non-minimized editor. If closed, ask the user to open Blender. If minimized/hidden, say Restore Blender from the taskbar or ask them to show it; maximization is unnecessary. If inspection is unavailable, report unknown and ask the user to make it visible. Pause attempts until resolved, then repeat the check. Do not restore, focus or maximize it automatically.
Inspect actual captured pixels. Minimized official captures have returned black on the observed Windows workstation. A visible window is not a guarantee: report black/stale captures, request show/restore, and recheck before retrying. Never substitute old screenshots or a clean render as fresh editor evidence. Saved render output does not need this editor prerequisite.
For supplied .snagx, inspect ZIP entries, extract/view the original full-resolution image temporarily, and preserve the original archive. Screenshots prove capture-time state only. Keep multiple/concatenated paths distinct and do not claim access to other AI chats that were not available.
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.

