main.tex is an arXiv-oriented technical report. Its numerical macros and matrix table are generated directly from a completed study:
python examples/agentcert-bench-24-task-study/render_paper.py `
--analysis examples/agentcert-bench-24-task-study/results-v0.5/analysis.json `
--paper-dir papers/agentcert-bench-v0.5
Run that command from the repository root. Then compile with latexmk -pdf main.tex
from this directory. The report covers 480 task runs with five repetitions per
cell, run variance, paired effect sizes, repetition-cluster sign-flip tests, and
Holm correction.
Do not submit the draft until an independent clean-clone run,
artifact redaction review, author metadata review, and claim review are complete.