git:20260813.2328887 to git:20260813.2b114ab

7 added, 4 removed. Audit A to A.

# Contents
- - Source captures live in one descriptive directory per capture, with Markdown, provenance, and any local assets kept together.
+ - Web captures live in one descriptive directory per capture with Markdown, `capture.json`, optional tool-owned `url-metadata.json`, inert evidence, and local `assets/`.
+ - PDF captures keep the original document, page-provenance text, image metadata, OCR evidence, and visual assets together in their capture directory.
# Guidelines
- - Keep each capture self-contained and record its original source, capture date, completeness, and warnings.
+ - Use `$save-url-kb` or `$save-pdf-kb` so source metadata, relative links, images, and provenance are handled consistently.
+ - Keep each capture self-contained and record its original source, capture date, completeness, expected-versus-captured counts, and warnings in `capture.json`.
+ - Keep search-derived candidates and provider attempts in tool-owned `url-metadata.json`; do not merge them into provenance or edit the sidecar by hand.
- Keep local assets beside their capture and use relative links from the capture Markdown.
- - Do not commit cookies, credentials, browser profiles, raw authenticated DOM, HAR files, or other session state.
+ - Do not commit cookies, credentials, browser profiles, raw authenticated DOM, HAR files, or other session state. A sanitizer does not make screenshots or authenticated prose safe to publish.
- Preserve captured wording and source authority. Put personal analysis in a separate linked note or a clearly labeled annotation.
- - Never present a partial capture as complete.
+ - Never present a partial capture as complete or change a capture's completeness by hand.