ob-methods-results-checklisted

# OB Methods Results Audit Review Methods and Results as a pre-submission audit. Follow the user's language unless the user requests another language. ## Core stance - Prioritize issues that may change core conclusions. - Treat the manuscript, tables, figures, outputs, scripts, and data as one result chain. - Use scripts for deterministic recalculation only. Use LLM judgment for design, construct, and inferential evaluation. - Do not claim that an analysis is correct without the necessary raw output or data. - Keep high-stakes decisions with the researcher. ## Portable runtime preflight Before any bundled command, resolve the directory containing the loaded `SKILL.md` to an absolute path, not relative to the user's current working directory. Either substitute that absolute path directly in each command or initialize: ```bash SKILL_DIR="/absolute/path/to/ob-methods-results-check" ``` Before auditing, check the runtime: 1. First, determine whether `python3` is available. 2. Only when Python is available, run: ```bash python3 "$SKILL_DIR/scripts/check_environment.py" ``` 3. If Python is unavailable, continue with a Markdown-only audit. Disclose that deterministic recalculation and HTML rendering were unavailable. 4. If Python or a bundled helper command fails, preserve the audit, disclose the failure, and continue with manual evidence review where feasible. Use the available capabilities and disclose any limitation in the report: - Treat `pdftotext` and `pdftoppm` as