# How BriefAuthority works — the specification

**The short answer:** BriefAuthority (in development) is built to read the final PDF in your browser, find each authority with its printed page numbers, show each one for you to confirm, and write a formatted .docx — without the brief leaving your computer. Specification as designed; confirmed at launch.

## Steps
1. Open the final PDF (text-based; scans need OCR).
2. Map printed pages (front-matter roman numerals, where page 1 starts).
3. Detect authorities: cases (full, short form, Id., supra), U.S.C./state statutes, C.F.R., court rules, constitutional provisions.
4. Review every entry and its pages; merge, link, re-categorize, exclude.
5. Export a .docx Table of Authorities with tab leaders.

## Specification
| Item | Detail |
| --- | --- |
| Where it runs | In the browser tab; the PDF is not uploaded |
| Platforms | macOS, Windows, Linux, ChromeOS; current Safari, Chrome, Edge, Firefox |
| Page references | Printed pages with front-matter offset; ranges for consecutive pages |
| Passim | Off by default; optional where the court allows it (Supreme Court Rule 34.2 says not to use it) |
| Output | .docx with categories, alphabetized entries, dot leaders |
| Not included | Citation validation (not a citator), Bluebook form checking, editing the brief |
| Price (planned) | $39 one-time or $9 per brief |

## Internal test
Greer v. United States (No. 19-8709): 70 of 70 cases found, exact page lists 84%. Bouarfa v. Mayorkas (No. 23-583): 59 of 61 cases, exact page lists 77%. Misses were mostly name-only short references, surfaced on the review screen. Statutes and rules not benchmarked. Internal, pre-launch.

## Privacy architecture
The PDF is opened with the browser File API, parsed by PDF.js, indexed by a JavaScript citation parser and written to a .docx in memory. It is designed with no upload endpoint for the brief; verify in the Network panel or offline.