Install
Optional ember swarm renderer for Sprite: a breathing ring that performs Agent-written scenes
For a published npm release:
npm install @fungi.computer/sprite-swarmpnpm, Yarn and Bun work the same way.
Entrypoints
Section titled “Entrypoints”@fungi.computer/sprite-swarm@fungi.computer/sprite-swarm/ember@fungi.computer/sprite-swarm/scene@fungi.computer/sprite-swarm/testing
What comes with it
Section titled “What comes with it”Installing @fungi.computer/sprite-swarm also installs @fungi.computer/art-library, @fungi.computer/sprite, effect, three, zod.
Install or set up
Section titled “Install or set up”npm install @fungi.computer/sprite-swarmThe package ships ESM only and depends on three.js, Effect, Zod and
Sprite at runtime. Add it only to a host that already
uses Sprite. The parent Sprite host still owns the player, its pixel limit and
its asset policy. The root entrypoint exports the swarm renderer and
builtinCharacters. ./ember is the ready-to-load local character with its
document and renderer. ./scene is pure: the scene schema and decoder, the
timeline helpers and the performance and progress schemas, with no three.js.
./testing holds the sample scenes the package’s own tests and its hosts’ tests
share; it is not part of the renderer.
For model elements, pass a host-owned model map as performanceOf’s fourth
argument: each catalog ID maps to its shipped GLB URL and height in metres.
Primitive-only scenes pass {}. The Sprite asset policy admits the GLB before
three.js decodes it; models must be self-contained. Missing or refused models
remain in the halo and their element names appear in progress’s unavailable.