Vibe-code Fusion 360 add-ins.
Give your AI agent deep knowledge of Fusion add-ins that use an HTML and JavaScript architecture for maximum UI flexibility. A complete, self-contained kit — no other purchase required.
One-time · self-contained · works with Claude, Codex, Cursor, Copilot & Gemini.
The problem
Fusion's add-in API is thinly documented.
The docs are thin and the palette architecture isn't obvious, so a general-purpose agent guesses at the API — and gets it wrong. You lose days to gotchas nobody wrote down.
The bridge is easy to get wrong
Palette architecture, event wiring, the HTML/JS-to-Python bridge — subtly wrong in ways that only show up at runtime.
The agent guesses
Without Fusion-aware steering, the AI guesses at API names and patterns — confidently, and wrong.
Every add-in starts from zero
No scaffolder, no templates, no conventions — just a blank folder and a docs site that doesn't quite answer the question.
Why listen to me
I build these add-ins — and taught my agent how.
I ship real Fusion tooling through the same thinly-documented API you're staring at. This kit is the steering I use to make my own agent good at it — packaged so yours starts out the same way.
The plan
Three steps to add-ins that ship
Drop in the bootstrap
Add the scaffolder, templates, and steering docs to your add-in folder.
Your agent learns Fusion
It picks up the API gotchas, palette patterns, and conventions automatically.
Build & test real add-ins
Ship add-ins that hold up — and a workflow you can repeat for the next one.
Inside the kit
6 items · 5 free · 1 with purchase
Skills your agent learns
Reusable commands your agent already knows how to run. Type one by name; the ones that only look and report also kick in when you ask in plain language.
/jgl-fusion-checkRuns every check a Fusion 360 React-palette add-in has — webapp audit, type check, unit tests and build; Python audit and unit tests; and the integration-test hash gate, refreshing it through the Fusion MCP server when stale — and reports gates and drift signals with an explicit SKIPPED line for anything not installed./jgl-fusion-commitStages a Fusion add-in's relevant changes and makes a local commit with a specific message and no AI attribution, checking that integration results travel with the source they were generated against; never pushes. On a long session it also refreshes docs/PROGRESS.md./jgl-fusion-mcpRegisters the Autodesk Fusion MCP server and proves it's live — registered, Fusion running with the add-in loaded, the right document open — before any MCP-dependent work such as headless integration tests, API probing, or viewport screenshots./jgl-fusion-pushPushes a Fusion add-in's committed work to the remote after checking for a stale integration-test hash and an out-of-date webapp build, asking first before any push to main or master, and never forcing or skipping hooks./jgl-fusion-review-codeReviews a Fusion add-in's whole codebase for structural problems and Fusion API misuse — duplication, multi-responsibility files, dead code, logic tangled into event handlers or components, type gaps, dependency health, stored entity objects, collection iteration, unregistered handlers — and produces a prioritized refactor plan if one is warranted./jgl-fusion-review-testingAudits a Fusion add-in's tests across all three layers — Python unit, TypeScript unit, and in-Fusion integration — for weak assertions, untested logic, tests that assert the implementation, mock_adsk gaps, stale integration hashes, and CI gates that can't fail, and produces a prioritized testing plan if one is warranted./jgl-fusion-scaffoldScaffolds a new Fusion 360 React-palette add-in by running the kit's tested scaffold.py generator from four inputs — add-in name, company, target workspace, description — then installs and builds the webapp, smoke-tests it, switches on the git hooks, and explains how to load it in Fusion./jgl-fusion-scopeRuns a 15–20 minute guided interview for a new Fusion 360 add-in — what it does, which workspaces it touches, whether it reads or mutates the model, how it persists state, and the MCP dev loop — then writes docs/SCOPE.md, adds a Project context section to AGENTS.md, and hands the four scaffold inputs to /jgl-fusion-scaffold./jgl-fusion-setup-testingWires the CI-runnable test layers into an existing Fusion add-in that lacks them — pytest with mock_adsk for Python, Vitest for the webapp, the CI workflow, and a pre-push hook — and explains the in-Fusion integration layer./jgl-fusion-statusGives a read-only session checkpoint for a Fusion add-in — what's committed, staged and in progress, integration-test hash freshness, webapp build freshness, anything risky left uncommitted, and whether enough has been generated since the last commit to checkpoint.
Wait, what's a "skill"?
From Napoleon Dynamite (2004): “I don't even have any good skills. You know like num-chuk skills, bow hunting skills, computer hacking skills. Girls only want boyfriends who have great skills!”
Modern AI coding tools let you teach your agent skills — named playbooks it can recognize and follow. Each card above is one your agent gets when you install the kit.
What's new in this version
- One rules file. Every major tool reads
AGENTS.md; Claude Code imports it. The architecture reference is looked up when a task needs it instead of loading on every turn. - Add-ins carry their own rules. The scaffolder copies the rules and skills into every add-in it creates, plus a commit hook that stops AI attribution in any tool.
- Leaner skills. Trimmed for current models, and the ones that change things run only when you ask.
Upgrading from an earlier download? Delete the old GEMINI.md, Copilot instructions file, and .cursor/rules/ — those tools read AGENTS.md now, and the old copies would load the same rules twice.
What's at stake
The difference Fusion-aware steering makes
With a generic agent
- Invented API calls that don't exist
- Days lost to undocumented gotchas
- A blank folder every single time
- Add-ins that break on the next update
With the Fusion kit
- An agent that knows the real API
- A scaffolder + templates to start fast
- Good dev practice baked in
- Add-ins you can test and trust
Get started
A self-contained Fusion kit.
A complete, self-contained kit — steering, scaffolder, templates, and skills. No other purchase required to start building.
Also building web or backend apps? The Vibe Code Starter Kit applies the same playbook there.
Pairs well with
Also working on web/backend projects? The Vibe Code Starter Kit applies the same playbook to that domain — fully self-contained, not a prerequisite for the Fusion kit.