---
title: "Review tools for verified weighting"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Review tools for verified weighting}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
```

WFC review tools explain feasibility and weight distortion without changing the
verified target.

```{r review, eval=FALSE}
precheck <- wf_precheck(
  design$data,
  target,
  id = design$roles$id
)

trim_advice <- wf_auto_trim(
  design,
  target,
  caps = c(2, 4, 6, 8),
  max_deff = 6,
  max_residual = 0.02
)
```

`wf_auto_trim()` returns a frontier and recommendation. It does not apply a cap
to a production result. A reviewed plan must contain the final bounds.

After weights are locked, decision readers receive a compact status and next
actions; statistical readers receive full diagnostics and provenance:

```{r reports, eval=FALSE}
wf_report(locked, audience = "decision")
wf_report(locked, audience = "statistician")
```

Both views are generated from the same object.