Connect an agent to PINOC
Give Claude, Cursor, or any MCP-capable agent direct tool access to your PINOC library.
PINOC's MCP server exposes the same motion library, character generation, and credit balance as the app itself, as tools an agent can call on its own. Connecting opens a Google sign-in in your browser. There is no API key to copy.
New to MCP? Read the protocol spec
Paste this into any agent that can add an MCP server from a prompt:
Connect to the PINOC MCP server at https://pinoc.mcp.viggle.ai/mcp and complete the Google sign-in it opens.What the agent can do
Free
list_motion_library
Browse the curated motion library. Free and instant.
take_library_motion
Take one named library clip. Free, instant, no credits.
take_library_motions
Take several named clips in one call, up to 25. It reads the catalogue once instead of once per name.
wait_for_tasks
Poll one or more generation tasks. Returns within 45 seconds by default, 90 at most, with whatever has happened by then.
get_asset
Look up one asset's metadata and signed download URLs, without downloading anything.
get_assets
Look up several assets by id in one call, up to 25, fetched in parallel.
download_asset
Write an asset straight to a local path. Only over a local connection; a remote server cannot reach your disk, so fetch the signed URL from get_asset instead.
get_credit_balance
The credit balance and what each paid tool costs. Free and instant.
Costs credits
generate_motion
Generate a new animation from a text description. 1 credit per second, 3 to 60 seconds, and you get 4 samples of the prompt for that one price. Returns a task id to poll.
generate_character
Generate a rigged 3D character from either 1-5 photos of one person or a text description. 6 credits, 10 with the refine pass. Returns a task id to poll.
The free tools need only the pinoc:read scope; the two that generate need pinoc:generate as well. A connection granted only the first is refused them before anything is submitted, so a read-only agent cannot spend credits.
Questions
Still stuck?
The Viggle Discord has a channel for PINOC, and the people building on it answer there.
Join the DiscordWhat is MCP?
The Model Context Protocol is an open standard that lets an AI agent call a service's tools directly, instead of you copying results back and forth by hand.
Do I need an API key?
No. Connecting opens a Google sign-in in your browser, and your agent registers itself with PINOC automatically.
Which agents can connect?
Any MCP-capable client that speaks the protocol over HTTP, including Claude and Cursor.
Does this cost credits?
The motion library, asset lookups, your balance and downloads are all free. A generated motion is 1 credit per second of clip and returns 4 samples for that one price; a character is 6 credits, or 10 with the refine pass. A task that fails refunds itself.
How long does a generation take?
Minutes. Nothing blocks: a generation returns a task id straight away and the agent polls it, which answers within 45 seconds by default and 90 at most. Several rounds of polling is the normal shape, and a task still running is never a reason to start a second one.
What permissions does it ask for?
Two scopes. pinoc:read covers everything that does not spend credits, and pinoc:generate covers the tools that do. You can grant just the first, or both, on the sign-in screen.
What do I do with the files?
Motions come as GLB, FBX and USDZ. GLB and FBX open in Unity, Blender, Unreal or any glTF tool with nothing extra installed; the skeleton is a 441-bone MetaHuman armature with no mesh, so it needs a character to drive. Characters are Gaussian splats in the .vsplat format, which is not a mesh and which no glTF or FBX importer opens. They render through the Viggle Splat Engine, as a Unity package or an npm package for the web. Get the Splat Engine
I already use Blender MCP or Unity MCP. How does this fit?
Those let an agent drive the software you already have open. This one gives it something to put in there: rigged characters and motion clips that export as GLB and FBX. They are complementary, and an agent connected to both can generate a character here and place it in your scene there.
How is this different from Mixamo?
Mixamo is a fixed library you browse by hand. PINOC has a curated library too, but an agent can also generate a motion that is not in it, from a written description, and rig your own model. The comparison page goes through where each one ends. PINOC vs Mixamo
The capabilities behind the tools
Text to 3D motion
What generate_motion runs on. Describe a movement, get animation data.
Image to 3D character
What generate_character runs on. Photos in, a rigged character out.
PINOC
The app itself, where every asset an agent makes can be previewed and edited.
Viggle API
The same jobs over REST, when you are building software rather than driving an agent.
See what people are building
Open the games portalGames made with Viggle characters live in the games portal. Play them.
