Expand treejs-arena prompt set to 25 scenes

[DogukanUrker]

What changed

treejs-arena now ships 25 frozen prompts instead of 20. The five new scenes cover capability axes the existing set did not touch: hard-surface product modeling, articulated motion, rigid-body collision, recursive procedural geometry, and planar reflection.

slug	prompt
ps5-gamepad	A PlayStation 5 DualSense controller, modeled in detail and slowly rotating under studio lighting.
hexapod-walk	A six-legged robot walking forward across a plane with a tripod gait, its feet staying planted while they carry the body.
domino-chain	A chain of dominoes falling along an S-curve, toppling one after another.
l-system-tree	A tree grown from an L-system, branching recursively and swaying in the wind.
reflective-pool	A rippling pool reflecting the sky and a few floating objects, seen from a low angle.

The prompts are appended, so existing slugs and TreeJSArena/<index> ids are unchanged and no shipped prompt was edited in place. PROMPT_SET_VERSION moves from v1 to v2 because the denominator behind a run's render rate changed, and runs should not be compared across that boundary. Scoring, artifact handling, and the gallery are untouched.

Wording follows the existing creative prompts: only hexapod-walk carries a spec clause ("feet staying planted"), because foot skating is the visible failure mode. The rest stay open-ended, since these scenes have no ground truth and an enumerated part list would just hand the model a checklist.

Benchmark or dataset files touched

• src/benchkit/benchmarks/treejs_arena.py — five prompts appended, PROMPT_SET_VERSION bumped to v2, class docstring no longer hardcodes a count.
• src/benchkit/benchmarks/__init__.py — registry description no longer says "twenty".
• tests/test_treejs_arena.py — prompt-set count assertions updated to 25.
• README.md — benchmark table count 20 → 25.
• AGENTS.md — creative rendering section no longer hardcodes a count.

No dataset files changed; this suite generates its prompts in code.

Validation

$ uv run ruff check .
All checks passed!

$ uv run pytest -q
................................................................ [ 21%]
................................................................... [ 43%]
.............................................................. [ 63%]
........................................................................ [ 87%]
......................................                                   [100%]

$ uv run benchkit --list | grep treejs
  treejs-arena   │        25 │ frozen         │               │ creative ·

The render path itself was not exercised here — this container has no browser, and the new prompts reuse the existing rendering and scoring code unchanged.

Notes

• Terminal or report snippet attached if CLI output changed — --list output above; the task count is the only CLI-visible change.

🤖 Generated with Claude Code

https://claude.ai/code/session_01BMNSQS8H71wZgxgfcapymT

---

Generated by Claude Code

Summary by CodeRabbit

• New Features

  • Expanded the Three.js Arena benchmark from 20 to 25 creative scene prompts.
  • Added scenarios featuring a PS5 gamepad, hexapod robot, domino chain, L-system tree, and reflective pool.

• Documentation

  • Updated benchmark listings and descriptions to reflect the expanded prompt set.
  • Clarified that the benchmark contains frozen creative prompts, without tying the description to a fixed prompt count.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 58c97426-e7cc-4ce1-ac2c-d04dccf99796

📥 Commits

Reviewing files that changed from the base of the PR and between dcf264a and c95e3b4.

📒 Files selected for processing (5)

• AGENTS.md
• README.md
• src/benchkit/benchmarks/__init__.py
• src/benchkit/benchmarks/treejs_arena.py
• tests/test_treejs_arena.py

---

📝 Walkthrough

Walkthrough

The Three.js Arena prompt set advances from v1 to v2. Five prompts are added, increasing the validated benchmark size from 20 to 25. Related descriptions and documentation no longer hard-code the former prompt count.

Changes

Three.js Arena prompt expansion

Layer / File(s)	Summary
Prompt set v2 src/benchkit/benchmarks/treejs_arena.py	The prompt set advances to v2 and adds five scenes: PS5 gamepad, hexapod, domino chain, L-system tree, and reflective pool. The class description removes the fixed count.
Count and validation updates src/benchkit/benchmarks/__init__.py, AGENTS.md, README.md, tests/test_treejs_arena.py	Benchmark descriptions remove the former count. The README lists 25 tasks. Tests validate 25 tasks, prompts, slugs, scenes, and task IDs through TreeJSArena/24.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: claude

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch claude/ps5-gamepad-3d-prompt-tgwx6q

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}