Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add sosadly/blockbench-mcp --skill blockbench-mcp-modelinggit clone --depth 1 https://github.com/sosadly/blockbench-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sosadly/blockbench-mcp/blockbench-mcp-modeling)<a href="https://agentmods.dev/skills/sosadly/blockbench-mcp/blockbench-mcp-modeling"><img src="https://agentmods.dev/badge/skills/sosadly/blockbench-mcp/blockbench-mcp-modeling.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00117 | $0.01788 |
| Opus 5 | $0.00059 | $0.00894 |
| Sonnet 5 | $0.00023 | $0.00358 |
| Haiku 4.5 | $0.00012 | $0.00179 |
Grade A, and why
blockbench-mcp-modeling scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Blockbench MCP Modeling Operator Guide
This skill makes you good at driving Blockbench through the BlockbenchMCP bridge
(mcp__blockbench__* tools). It captures a workflow and a set of scripts that have been
verified live against Blockbench 5.1.4 + GeckoLib. Follow it and models come out detailed,
smoothly textured and close to the reference — not a pile of flat boxes.
Connection check (do this first)
The tools talk to a local server inside Blockbench. If a tool returns
Cannot reach Blockbench on http://127.0.0.1:8787:
- The desktop Blockbench app must be open and the MCP server started: in Blockbench,
Tools ▸ Start MCP Server. The server stops when Blockbench is closed. - Then call
get_statusto confirm. If a model is requested, also callget_guideonce.
The golden workflow (always loop)
get_status / get_guide
-> plan bones (sketch the skeleton + joint pivots)
-> add_groups (bulk: whole posed bone hierarchy in one call)
-> add_cubes (bulk: 20-100 cubes; segment limbs; mirror L/R in the same call)
-> PACK BOX UV (execute_script shelf packer — see references/workflow-and-scripts.md)
-> create_texture (mid-tone fill, size 64-192 depending on cube count)
-> BAKE TEXTURE (execute_script smooth bake — gradients + soft mottle + per-island blur)
-> paint features (eyes, nose, claws, runes, glow cores)
-> screenshot_views (front / side / back / iso AND the reference's exact angle)
-> check_model (fix untextured faces, bad UVs, unparented cubes)
-> COMPARE to the reference, fix what is wrong, screenshot again. Repeat 2-4x.
-> save_project + export (geo.json, texture png, animation json)
Iterate at least 2-3 passes. The first pass is never good enough.
Hard rules (these are the difference between good and bad)
- Box UV does NOT auto-pack. Newly created cubes all get
uv_offset [0,0]and overlap. You MUST run the shelf-packer script (setsuv_offset+ callscube.mapAutoUV()) before texturing, or every face shares the top-left pixels. Re-pack whenever you add/resize cubes.
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 110 lines · 117 tokens per session scan A d07b039e0380
blockbench-mcp-modeling is a skill published in the GitHub repository sosadly/blockbench-mcp (5 stars, last pushed 2mo ago), licensed MIT. It adds 117 tokens to every session and 1,788 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
pipeline-automation
Automate 3D asset pipelines, including batch optimization, material baking, and exports for platforms like Unity and VRChat.
ashfox
Create, edit, review, animate, and export game-ready low-poly assets in the ashfox web workbench through an AI agent. Use for low-poly modeling, deterministic textures and UVs, rigs, idle or motion animation, GLB, Bedrock, or GeckoLib exports, and Blockbench-free asset production.
blender-modeling
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…
reference-to-3d
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…
blender-uv-texturing
UV unwrap, atlas-map, project textures, use alpha decals, bake maps/lightmaps, and prepare texture-driven Blender assets for glTF/GLB export. Use whenever the user provides texture packs, texture atlases, decals, UV layouts, lightmaps, wants a texture to fit a mesh 1:1, or reports stretched/off textures. Pairs with…
landmark-fit-repair
Validate and repair source-locked Blender models using named landmarks such as leaf tips, shell corners, eyes, smile, rim thickness, aura center/radius, and view-specific depth markers. Use when bbox/IoU is insufficient and the model must align to templates at designed feature points.