worldjen-score

name

worldjen-score

description

Score an individual generated clip, video, or image. Upload one (or a few) media files to the user-scoped Score session and get raw per-dimension quality scores back, with optional custom prompt and dimension selection. This is the DEFAULT for "score it / score this / score these" whenever the target is a piece of media — a clip, video, image, shot, frame, generation, or output — and NOT an entire AI model. Use for quick feedback on a generation, A/B-ing two clips, or inspecting evaluator scores without a full benchmark. If it is unclear whether "it" means a single clip or a whole model, ask the user before scoring. Backed by `worldjen score get/reset/upload`. For ranking several clips that share a prompt use `worldjen-rank`; for benchmarking an entire model use `worldjen-bench`.

Preamble (run first)

Best-effort update check. Fails silently on network errors or non-marketplace installs. If output starts with UPGRADE_AVAILABLE or JUST_UPGRADED, surface it once to the user and continue with the skill workflow.

{
  for _p in \
    "$HOME/.claude/plugins/marketplaces/worldjen/bin/check-update" \
    "$HOME/.codex/.tmp/plugins/plugins/worldjen/bin/check-update"; do
    if [ -x "$_p" ]; then "$_p" 2>/dev/null && break; fi
  done
} 2>/dev/null || true

WorldJen — Score

"How good is this clip?" Score is the single-clip surface: upload a video (or image) and get raw per-dimension scores back. Each user has exactly one Score session; uploads accumulate in it until you reset.

There is no leaderboard or comparison here — just the dimension scores for the clips you uploaded. For comparative ranking of clips that share a prompt, use worldjen-rank. For a full benchmark across many prompts at scale, use worldjen-bench.

Resolve the target first

This skill scores media — a clip, video, or image. Before scoring, make sure you know what the user wants scored:

• If they name or attach a media file (or it's obvious from context), score it here.
• If they're asking to evaluate an entire AI model (a model id, checkpoint, Hugging Face repo, "the model", "my model"), stop and use worldjen-bench instead.
• If they explicitly want clips ranked / compared / sorted against each other for one prompt, use worldjen-rank.
• If the target of "it / this / these" is ambiguous — you can't tell whether they mean a single clip or a whole model — ask the user before doing anything. Do not guess.

Auth

Set WORLDJEN_API_KEY (get one at https://www.worldjen.com/settings/api-keys), or pass --api-key.

If worldjen is not installed yet, run pip install worldjen first (or use the worldjen-install skill).

Get the Score session

worldjen score get          # text view
worldjen score get --json   # machine-readable

Upload and score

Upload one or more clips. Add --wait to poll until scores land (or --timeout elapses; default poll interval 3s, tune with --poll-interval).

worldjen score upload clip.mp4 --prompt "a cat sitting on a windowsill" --wait

Pick specific dimensions (otherwise the default set is scored). List them first with worldjen dimensions list --json:

worldjen score upload clip.mp4 --dimensions aesthetic_quality,dynamic_degree --json

Images (t2i) work the same way:

worldjen score upload image.png --prompt "neon city, cyberpunk" --dimensions prompt_adherence --wait

Confirm before destructive operations

• worldjen score reset — clears all videos from the Score session; the session id stays

This cannot be undone. Confirm with the user before running, especially if they have been iterating recently.

SDK alternative

import worldjen

result = worldjen.score.upload("clip.mp4", prompt="a cat", wait=True)

Stop and ask when needed

• Missing WORLDJEN_API_KEY
• The user wants to "delete" the session — there's no delete, only reset (which keeps the session id)
• The user is unsure which dimensions to score against — list them with worldjen dimensions list --json

See also

• worldjen-install — install the SDK and CLI
• worldjen-rank — comparative ranking of clips that share a prompt
• worldjen-bench — whole-model benchmark across all dimensions, at scale
• worldjen-leaderboard — public leaderboard
• worldjen-runner — GPU worker host setup

For examples and troubleshooting, see references/examples.md.