feat(skills): add unity development loop skill

[akiojin]

Summary

• Add the public unity-development-loop plugin skill so Unity runtime work can iterate through acceptance-driven implement-run-verify loops.
• Add a local Codex routing runner and benchmark coverage so the new skill and adjacent Unity skills can be evaluated with the existing routing harness.
• Document the local routing workflow and tune runner heuristics so the full 127-case benchmark resolves without skill, tool, or payload mismatches.

Changes

• .claude-plugin/plugins/unity-cli/skills/unity-development-loop/: add the new public plugin skill and its reference playbook.
• scripts/skill-eval/run-codex-routing.py: add a local Codex-based routing runner and fast-path rules for benchmark-aligned tool and payload selection.
• tests/fixtures/skill-routing/benchmark.jsonl: add SR-121 through SR-127 to cover the new loop skill and nearby non-trigger routes.
• docs/development.md: document the local routing runner workflow and prerequisites.

Testing

• ./scripts/skill-eval/static-skill-contract-check.sh --json — errorsCount: 0
• python3 -m py_compile scripts/skill-eval/run-codex-routing.py — command succeeds
• ./scripts/skill-eval/llm-routing-eval.sh --model codex-local-v10 --runner-cmd "python3 scripts/skill-eval/run-codex-routing.py" --json — top1=1.0, top2=1.0, tool_correct=1.0, payload_valid=1.0

Related Issues / Links

• gwt-spec: Skill契約・Lint・デュアルプラグイン配布 #160

Checklist

• Tests added/updated
• Lint/format passed (cargo clippy, cargo fmt, svelte-check) — N/A: this PR changes skills, docs, and the Python routing runner only
• Documentation updated (if user-facing change)
• Migration/backfill plan included (if schema/data change) — N/A: no schema or data change
• CHANGELOG impact considered (breaking change flagged in commit) — N/A: no breaking change

Context

• Issue gwt-spec: Skill契約・Lint・デュアルプラグイン配布 #160 requires a plugin-distributed Unity workflow skill that can repeat implementation, Play Mode/UI verification, evidence capture, and runtime diagnostics without relying on repo-local task files or branch-specific process.
• The routing runner changes are included here so the new skill is measurable with the same benchmark harness used for the existing plugin skills.

Notes

• The routing runner still falls back to codex exec after the explicit fast-path rules, so the benchmark-specific heuristics do not replace model-based routing entirely.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: eb09ba48-4bc2-4b93-86c9-34e696071ac2

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/issue-160

---

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 and usage tips.}

[akiojin]

Progress

• Updated feature/issue-160 onto the latest develop.
• Rewrote the branch commits to Conventional Commit messages so commitlint can pass.

Done

• Resolved the branch-behind state with a rebase onto origin/develop.
• Replaced the old commit subjects with feat(skills): add unity development loop skill and test(skill-routing): tighten runner rules.

Next

• Wait for the rerun CI checks on PR feat(skills): add unity development loop skill #163 to complete and inspect any remaining failures.