Author a Mesh Animation Test Case

Scope

Scaffold an asset-generation test case for a rigged, animated meshed model: a model sculpts per-part signed-distance fields with the mc-anim, sn-anim, or dc-anim binary and rigs them. The case fixes the required animations; the parts and joints that realize them are the model’s to invent. Read Authoring a Mesh Animation Test Case for the full procedure; Voxel cases is the authoritative schema.

For a static meshed model see Author a Mesh Model Test Case. For a rigged cube model see Author a Voxel Animation Test Case.

Layout

A version lives at test-cases/asset-generation/<difficulty>/<slug>/<version>/. A version with runs recorded against it is frozen; revise a case by adding a new version.

test-cases/asset-generation/<difficulty>/<slug>/<version>/
  test-case.toml   # manifest: type, asset_kind, voxel, tool, output, model
  variants/        # one standalone TOML file per variant
  prompt.hbs       # rendered into the harness instruction; not seeded
  changelog.md     # required per-version entry; not seeded
  description.md   # site blurb; not seeded
  specs/brief.md   # what to sculpt and how the tool behaves; seeded

A run seeds the brief, <binary>.config.json, and a rig.json pre-populated with the required animation declarations alone, with empty tracks, parts: [], and joints: []. The -anim binary’s --help is the operation and rig-subcommand contract. Core emits the per-part meshes and the filled rig.json on render. The case declares no [[reference]] and carries no target model.

Steps

Pick a catalog slug and an articulated subject, then the algorithm: faceted mc-anim, smooth sn-anim, or crisp dc-anim. Only dc-anim exposes the per-primitive --sharp tag. The algorithm fixes both asset_kind and [tool].binary.
Fix the required animations: one [[model.animation]] per motion, each with a unique name such as march or radar_spin, a loop flag, and an auto_play flag, where true means a self-playing idle. Leave parts, joints, pivots, ranges, and pose angles to the model.
Write specs/brief.md: the subject, orientation, the volume framing, the exact opaque #rrggbb palette, how the binary meshes each part’s field with add-*, subtract-*, and --blend, the features that must read, and each animation’s behavior in prose. Keep the brief self-contained, and specify what to build rather than how.
Write prompt.hbs. It renders in strict mode against {{variant.*}}, {{#each specs}}, {{workspace}}, {{time_limit_hours}}, and {{voxel.*}}. Point the model at the brief and the binary’s --help, and require a render before finishing.
Write test-case.toml per the table below. [tool].preview and [output].actions each carry the {part} token, as in parts/{part}.png and parts/{part}.actions.json.

Declared	Rejected
`type = "asset-generation"` and `asset_kind` (`mc-animation` / `sn-animation` / `dc-animation`)	`[canvas]`, replaced by `[voxel]`
`changelog`; `[voxel]`; `[tool].binary` with a `{part}` `preview`; `[output].actions` with `{part}`	`[[reference]]`, since there is no target model
`[model]` carrying `[[model.animation]]` entries (`name`, `loop`, `auto_play`)	`[build]` and `[[check]]`, since there is no served build
`variants` as a root key, first entry the default; the single `overall` `[[domain]]`	`[[review_item]]`, since the rig is judged as a whole on one rating

The Aegis six-legged walking fortress is the worked example, rigged once per algorithm as aegis-mc-anim, aegis-sn-anim, and aegis-dc-anim. Read the one matching your surface.

Validate

Run these for every variant.

npm run lint:specs
tcab prompt --test-case <slug> --version <version> --variant <variant>
tcab seed   --test-case <slug> --version <version> --variant <variant>

prompt catches strict-mode template and manifest errors, including duplicate animation names and a missing {part} token. seed writes the seeded repository under tmp/, where you confirm the brief and the pre-seeded rig.json are self-contained. After editing, force a re-ingest so a backend-driven run picks up the change; see Running the Local Service Stack.

Next steps

Create a Mesh Animation Variant to add a brief variation.
Run a Test Case to exercise it end to end.
Review a Run to score the produced rig against the brief.