autodeskfusion

autodeskfusion is a skill for Claude Code, Codex from svd-ai-lab/sim-cli. It costs 79 tokens per session (1,449 once invoked), scanned A, original, Apache-2.0.

An automation guide for working with Autodesk Fusion, also called Fusion 360, through its local connection server. It covers inspecting a live Fusion desktop session, editing designs, checking models, exporting files, and running short Python code inside Fusion.

In plain words
What is it for?
It helps coding agents inspect documents, bodies, and sketches; change Design workspace models; verify exports; and run small in-application Fusion API scripts.
Why use it?
It prevents a common setup error: Fusion's `adsk` Python interface runs inside the Fusion application, so it should not be imported from ordinary host Python. It also helps choose between live Fusion calls and offline file processing.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit It helps coding agents inspect documents, bodies, and sketches; change Design workspace models; verify exports; and run small in-application Fusion API scripts.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/svd-ai-lab/sim-cli/autodeskfusion
Install

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.

Any agent
npx skills add svd-ai-lab/sim-cli --skill autodeskfusion
Clone the repo
git clone --depth 1 https://github.com/svd-ai-lab/sim-cli

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for autodeskfusion

README.md
[![agentmods](https://agentmods.dev/badge/skills/svd-ai-lab/sim-cli/autodeskfusion/github.svg)](https://agentmods.dev/skills/svd-ai-lab/sim-cli/autodeskfusion)
Your own site
<a href="https://agentmods.dev/skills/svd-ai-lab/sim-cli/autodeskfusion"><img src="https://agentmods.dev/badge/skills/svd-ai-lab/sim-cli/autodeskfusion/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for autodeskfusion

Your own site · 80×15
<a href="https://agentmods.dev/skills/svd-ai-lab/sim-cli/autodeskfusion"><img src="https://agentmods.dev/badge/skills/svd-ai-lab/sim-cli/autodeskfusion.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 79 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,449 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00079 $0.01449
Opus 5.5 $0.00032 $0.00580
Sonnet 5.5 $0.00016 $0.00290
Haiku 4.5 $0.00008 $0.00145

Measured 2d ago against content hash c3b96ab7b53b, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

autodeskfusion 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 2d 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.

plugins/autodeskfusion/src/sim_plugin_autodeskfusion/_skills/autodeskfusion/SKILL.md · 139 lines

How it starts

The opening of the file, as written. The whole thing — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Autodesk Fusion via MCP

Use this skill when working with Autodesk Fusion through its official local MCP server. Fusion's adsk Python API runs inside the Fusion desktop process, so prefer short, inspectable MCP calls against the live desktop session.

Tool Choice

Use the narrowest useful primitive:

  • Direct MCP: best for exploration, tool/resource discovery, quick Fusion API probes, and one-off scripts when the agent can call the MCP endpoint itself.
  • sim connect/exec/inspect: use when session history, shared sim serve access, standard logs, or plugin discovery make the workflow easier to resume.
  • Host Python: use for saved file parsing, JSON/report post-processing, and acceptance checks that do not need Fusion's in-process adsk modules.

Do not import adsk from host Python. If a task needs adsk, execute that code in Fusion through MCP.

MCP Workflow

  1. Keep Fusion open and visible. Human-owned dialogs such as sign-in, account selection, cloud-save prompts, and first-run setup stay outside automation.
  2. Discover the endpoint. Prefer SIM_AUTODESKFUSION_MCP_URL when provided; otherwise try Autodesk's documented default and, if needed, inspect local Fusion-owned loopback listeners. http://127.0.0.1:55484/mcp is a common Fusion MCP URL observed on this machine, but do not hardcode it: the port can change across machines, Fusion versions, or local service conflicts.
  3. Call MCP initialize, then tools/list. Let the returned tool schemas guide the next call; do not assume a fixed port or permanent tool list.
  4. For Python execution, prefer the discovered execute-capable tool, commonly fusion_mcp_execute. Send scripts with a def run(_context: str) entrypoint unless the current tool schema says otherwise.
  5. After GUI-visible mutations, capture or request visual evidence before the next edit.

Stability Lessons

  • Treat Fusion as a strict CAD kernel, not a tolerant visualizer. Meshes, renderers, and preview tools may display geometry that Fusion rejects as an invalid B-rep, especially near self-intersections, tight curvature, tiny clearances, or very dense paths.
  • Increase complexity step by step. Start with the smallest inspectable primitive, then add bodies, repeats, detail, and export steps incrementally. After changing a path, size, topology, or API method, return to a small smoke case before rebuilding the full model.
  • Run cheap host-side gates before asking Fusion to create solids: bounds, units, entity counts, clearance checks, curvature or feature-size checks, and expected export targets. If a gate shows the requested solid is geometrically impossible or marginal, record the limitation instead of forcing the operation.
  • For near-contact geometry, record explicit contact diagnostics rather than relying on screenshots: nominal feature size, minimum relevant distance, clearance margin, and the sampled/contact pair. Treat intended touching as a small positive CAD margin when the downstream kernel is sensitive to exact tangency.
  • If a primary Fusion feature constructor fails, try a semantically equivalent native construction only when it preserves the same topology, dimensions, and source geometry. Record both the fallback method and the original failure in metadata so the fallback remains auditable.
  • Separate model creation from heavyweight export. When a full STEP/archive export risks freezing Fusion, first create the model with export disabled, capture review screenshots or diagnostic markers, then export smaller smoke cases or retry the full export only after the geometry is accepted.
  • Use short MCP/client timeouts for exploratory calls. If a step fails or times out, reduce scope or switch to a simpler representation; do not keep retrying heavy operations with longer timeouts.
  • Keep generated Fusion documents under control. Before creating another automation document, close or save prior generated documents as appropriate, and avoid accumulating multiple generated projects in the desktop session. Treat human-owned unsaved work conservatively.
  • Prefer scripts with explicit runtime options supplied by context or environment variables, such as geometry mode, detail level, sample count, export mode, output tag, and diagnostic simplifications. Do not bake local paths, ports, sizes, or temporary settings into reusable scripts.
  • If Fusion becomes unresponsive, stop sending more MCP calls. Inspect the process state, restart Fusion only when appropriate, then rediscover the MCP endpoint and resume from the smallest passing step.
  • Preserve metadata with every export: sources, units, coordinate mapping, build mode, simplifications, bounds, object names/counts, output files, and failure messages.

Read the full file on GitHub · 139 lines

Changes

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.

  1. 2d ago First seen · 139 lines · 79 tokens per session scan A c3b96ab7b53b

Subscribe to this mod's changes

autodeskfusion is a skill published in the GitHub repository svd-ai-lab/sim-cli (239 stars, last pushed yesterday), licensed Apache-2.0. It adds 79 tokens to every session and 1,449 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-19.

Related

Other skills, from other repositories

bdf_font_converter

Convert a Unicode BDF bitmap font in writable storage to the DFN1 format used by the display module.

espressif/esp-claw · 27 tokens

solidworks-vibecad

A planning guide for turning natural-language mechanical design requests into structured, parameterized SolidWorks plans that can be reviewed before execution.

wzyn20051216/solidworks-automation-skill · 96 tokens

fluidcad-assembly

Building multi-part models in FluidCAD: assemblies, part files, inserting and mating parts, connectors, exposed geometry, part parameters, sub-assemblies, replicated instances, STEP parts, project units, and assembly export. Use this skill whenever the user wants an assembly, a mechanism, a hinge, a slider, several…

Fluid-CAD/FluidCAD · 154 tokens

gen-assembly

Generate a PartCAD assembly (an ASSY that composes parts with placement and/or mates) from a description, writing down what it is required to be, generating or reusing the component parts and validating the result with the PartCAD CLI. Use for /pc:gen-assembly or when the user asks to generate or create an assembly…

partcad/partcad · 80 tokens

add-interfaces

Enrich an existing PartCAD part with connection interfaces and ports (mating metadata) so it can be mated to other parts automatically. Use for /pc:add-interfaces or when the user asks to add interfaces, ports, connectors, or mating information to a part, or to make parts snap/connect/assemble together.

partcad/partcad · 69 tokens

gen-part

Generate a PartCAD part from a natural-language description (and optional reference images/requirements) by writing down what it is required to be, authoring a CAD script, and validating it with the PartCAD CLI. Use for /pc:gen-part or when the user asks to generate, create, or model a single mechanical part.

partcad/partcad · 70 tokens