# Nathan AI — complete archive inventory
Snapshot: November 5, 2026. Additive module: existing repository files are preserved.

## Open
Serve this repository with a static HTTP server and open `archive-index/2026-11-05/index.html`. The explorer searches every listed entry, filters by type and paginates results.

## Included
- 5,964 files and 139 folders from all 31 pages of the recursive owned archive listing.
- 19,718,607,299 recorded bytes; this is metadata, not downloaded-byte verification.
- 346 repeated filename groups retained without assuming identical bytes.
- Complete inventory in JSON and CSV; exact source IDs and paths retained.
- Searchable browser explorer using text-only rendering.
- Read-only local inspection script: SHA-256 for every inspected original; ZIP member metadata; SQLite tables, columns and row counts; CSV rows; JSON root counts; text lines, characters and whitespace-delimited words; optional Pillow image structure, dimensions, frame counts and pixels per frame.

Source links require the user's own source access. Committing this index does not grant access to source files.

## Coverage
This snapshot is a file-level inventory, not all internal data points. ZIP payloads and nested archives, every database value, decoded pixel hashes, historical versions, deleted files, shared folders, external repositories and complete conversation exports are not comprehensively inspected. All entries remain candidates for deeper inspection. File count, index rows, words, chapters, books and unique completed works are distinct measures.

The 6,056 source rows in Totality v009 are a separate historical record measure; do not equate them to the 5,964 file identities in this snapshot.

## Expand with original bytes
On a local computer, use an owner-selected directory of originals. The script does not download or execute them.

```sh
python3 inspect_originals.py /absolute/path/to/originals --output /absolute/path/outside-originals/inspection.jsonl
```

Pillow is optional for image inspection. Each JSONL row records measured findings and any error. ZIPs are listed without extraction, decompression or execution. Encrypted archives are identified, not bypassed. SQLite connections use read-only mode. Symlinks are skipped. Output must be outside the inspected root. A SHA-256 hash measures file bytes; it does not establish authorship, truth, or functional readiness.

Keep deeper inspection output owner-controlled: database schemas, paths and source descriptions can contain personal information. Do not commit credentials, passwords, private keys or seed phrases.

## Verification performed
Inventory: 6,103 entries and 5,964 distinct file IDs. Inspector fixtures: text word counts, SQLite row counts, ZIP member sizes and SHA-256 generation. Explorer JavaScript syntax checked.


## Subject catalog and B / 8 discovery crosswalk

The [additive subject catalog](../loc-b8/index.html) uses the existing owner Library of Tabor shelving file. It preserves 19,317 source items, links 5,851 discovery candidates without creating duplicate source copies, and keeps original completion status and work remaining visible. [Read the integration and coverage notes](../loc-b8/README.md). Private originals and raw recovery records remain in the owner-controlled archive.
