Everything stays
unboxed.
No forced 14 boxes. No collapsed manuscripts. No guessed completion. Every recovered source retains its identity, original title, path, status, version reference and raw record when available. One original may appear in many searches without becoming many duplicate originals.
Different sources and count units overlap. The owner catalog's 19,317 underlying private originals are not included as files in the public GitHub repository. This screen loads public indexes and any originals you explicitly select, without uploading them.
Your open recovery table
Choose a source collection to load. No private archive is accessed automatically.
Bring in the originals — privately
Choose your owner catalog or recovery files. Accepts JSON, JSONL, CSV, TXT, Markdown, and source-code text; directory selection keeps individual file paths. Files stay in this browser tab. This does not publish, modify, overwrite, or transmit them.
ZIP and SQLite are indexed as original files, not unpacked or executed by this browser. Use the existing read-only archive inspector for member/row inspection. File sizes over 15 MB are indexed by name only; no silent truncation masquerades as full recovery. No local records are uploaded or placed into browser storage.