メニュー
Build the paper + arXiv submission bundle. Use before arXiv upload or when preparing a workshop submission. Runs latexmk, produces paper/arxiv-submission.tar.gz, then validates all result tables against committed benchmarks/*/results.md.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Run the pre-publish validation gate for a new release. Use before `make publish` or Zenodo deposit. Takes a version argument, runs make release-check, validates CHANGELOG [Unreleased] is non-empty, verifies version strings align, and previews the Zenodo metadata sync.
Run one of the project's comparative benchmark studies (realistic, ablation, appraisal, comparative). Use when executing Pre-reg S2/S3 or running appraisal quality. Validates env, runs the make target, summarizes results, and prompts for evidence update.
Run the LoCoMo external benchmark (Pre-reg Study S1) end-to-end. Use when executing or re-running the AFT vs naive_rag comparison on LoCoMo10. Enforces dry-run-first, env-var check, background execution, and post-run evidence update.
Update docs/research/09_current_evidence.md after a benchmark study completes. Use after any bench-locomo, bench-realistic, bench-ablation, or bench-appraisal run. Pulls numbers from results.md, enforces DISCOVERY vs CONFIRMATORY framing, cross-references the pre-reg.
Verify pre-registration integrity — no uncommitted edits to benchmarks/preregistration.md, and declared studies match produced result files. Use before paper building, release gates, or when resuming scientific work after a break.
| name | paper-bundle |
| description | Build the paper + arXiv submission bundle. Use before arXiv upload or when preparing a workshop submission. Runs latexmk, produces paper/arxiv-submission.tar.gz, then validates all result tables against committed benchmarks/*/results.md. |
| disable-model-invocation | true |
| allowed-tools | Bash(make paper*) Bash(tar *) Bash(git status*) Bash(git log*) Read |
Builds paper/arxiv-submission.tar.gz from paper/main.tex and validates
that all numeric tables in the paper match committed results files.
Run /prereg-guard first. If it reports any FAIL, resolve before building.
make paper-arxiv
Expected: exit code 0 and paper/arxiv-submission.tar.gz created.
On latexmk failure: read paper/main.log for the first error line.
Common causes: missing .bbl (run pdflatex + bibtex + pdflatex × 2
manually), undefined citation, or broken figure path.
tar -tzf paper/arxiv-submission.tar.gz
Must contain: main.tex, main.bbl, refs.bib, figures/, tables/.
Missing main.bbl → bibliography not compiled.
Read each file in paper/tables/ and compare numeric values against:
| Table source | Results file |
|---|---|
| LoCoMo F1 / judge_acc | benchmarks/locomo/results.md |
| Realistic top1_acc + CI | benchmarks/realistic/results.md |
| Ablation Δ, d, p_adj | benchmarks/ablation/results.md |
| Appraisal pass rate | benchmarks/appraisal_quality/results.md |
Flag any mismatch between the table value and the current results file. Stale numbers must be updated before submission.
Read paper/main.tex §Limitations. Confirm it mentions:
git status --short benchmarks/
All benchmarks/*/results.{json,md} files referenced in the paper must be
committed. Uncommitted results → commit them before submitting to arXiv.
Output a checklist:
✓ / ✗ Build: exit code 0, arxiv-submission.tar.gz created
✓ / ✗ Archive: main.tex, main.bbl, refs.bib, figures/, tables/ all present
✓ / ✗ Tables consistent with results files (list any mismatches)
✓ / ✗ §Limitations mentions DISCOVERY/CONFIRMATORY and null ablation
✓ / ✗ All referenced results files committed