Visual Keybook

1. Quick start

Use Automatic to run the applicable tests from one set of known data. Use the other tabs to inspect a result or control a single experiment. Calculations run on your computer in the browser; an open site does not provide a remote search service.

1. Open Automatic and choose Load demo.
2. Leave the sample range and budgets in place, then select Run all selected tests.
3. The known demo scalar is 424242. Its verified position is page 425, line 242. The report identifies it as supplied data.
4. Expand a test’s Measurements and exact values to inspect what was computed. Some tests will be skipped because the demo has no suitable data for them.
5. Select Export report .txt for reading or Export report .json for the structured results and reusable input setup.

For your own experiment, replace the demo data with one target and its known range. Leave unknown fields blank. Clear the demo scalar before entering an unrelated address or public key.

2. Inputs and number formats

Check the notation before running. In Automatic and Anchors, plain digits are decimal and hexadecimal numbers need an explicit 0x prefix. In Keybook, Pair Lab, CPU Lab and the Wormhole search, the relevant notation selector controls how numbers are read.

All supplied fields must describe the same target. The program checks agreement between the scalar, public point, address, page/line and known digits. A known scalar must lie inside your stated range.

Paste long integers as text. Avoid scientific notation, rounded spreadsheet values, separators and arithmetic expressions. A WIF is an output of the known-scalar derivation; it is not an accepted Automatic target or scalar input.

Automatic accepts the single-key address templates implemented in CPU Lab. P2SH comparisons assume nested P2SH-P2WPKH, and Taproot comparisons assume a key-only output with no script tree. Unsupported scripts cannot be identified from their address alone. Address columns produced by the derivation test are labelled as mainnet outputs.

3. Automatic workflow

1. Enter known data. Give the run a name. Paste the target, then open Additional known data only for facts you possess. Use current key copies the scalar and point currently selected in Keybook.
2. Set the range. Enter the lower and upper scalar. A smaller stated range does not become more likely merely because it is small.
3. Set work limits. The main search budget applies separately to each search method. Open All test budgets for sampling, benchmark and table limits.
4. Choose tests. All 18 are selected initially. Validation always runs. Tests keep their fixed order; a disabled prerequisite may cause a dependent test to be skipped.
5. Run. Automatic takes a snapshot of the current atlas, saved coordinate index and solved rows already imported into CPU Lab. It then runs one selected test at a time.
6. Inspect and export. Read each status and its details. Use Open in book for a verified result, or Inspect candidate for an explicitly unverified estimate.

Pause waits for the next checkpoint and keeps the current worker in this tab. Resume continues it. Stop ends the sequence and retains completed results and available partial counters. A stopped sequence has no resume control; starting again creates a fresh run. Inputs stay locked while running or paused.

Switching tabs is not the same as pausing. Keep the browser open while a run is active. Browser throttling or computer sleep can delay progress. A new Automatic run replaces the displayed report, so export anything you want to keep first.

Sampling, walk and test seeds make those parts repeatable with the same inputs and references. Timing still varies. Load setup from report restores inputs for a new run using the current references; it neither resumes a closed worker nor accepts imported results as verified.

4. The 18 Automatic tests

A public point or scalar found by exact recorded-data lookup can unlock later tests. Each row below describes one step; the result panel reports whether it ran, was skipped, stopped, or failed.

5. Reading results

Counts from different methods can overlap. Do not add sampled scalars, pattern checks, oracle checks, walk steps and giant lookups together as unique coverage. Even an exhaustively checked interval only rules out a match within its stated bounds and constraints.

Editing Automatic inputs after a run marks its report stale and disables result navigation. Exports still describe the earlier input snapshot until you run again.

6. Keybook and exact navigation

Choose Notation, enter a valid private scalar and select Derive chain. The six cards display the scalar, compressed WIF, compressed SEC, X coordinate, HASH160 and P2PKH address. The difference panels compare their QR projections. Use − 1, + 1 or Mirror n−k to move from the current scalar.

The book contains 1,000 scalar positions per page. Page and line start at 1; the final page has 336 valid lines. Moving to a page does not imply that an unknown target is on that page.

1. In Navigate from a reference anchor, enter an exact scalar using decimal or explicit 0x, then select Set anchor. You can also choose Use current scalar.
2. Enter a signed scalar offset d and select Go to offset. The destination is k = a + d. Offsets are scalars, not pages.
3. Alternatively enter a page and line and choose Open page and line, or use the configured Step size.
4. Flip book side changes the reference to n − a and the position to n − k; the offset changes sign. Flipping twice returns to the starting position.

Navigation stops at the valid scalar boundaries rather than wrapping. A navigation anchor is a known reference scalar; it need not have the special natural QR-mask signature.

7. Natural optimum hits and the 3D book

In the Keybook tab, scroll to Find natural optimums. Set the center notation and scalar, a radius on each side, and the worker count. The radius is a decimal count of scalars. The interval is clipped at 1 and n − 1; a radius of 500 normally gives 1,001 candidates.

1. Choose Scan full interval or Stop at first optimum, then Start new scan.
2. Use Pause and Resume for the current scan. Worker count controls local browser workers, not external computers.
3. Open View all optimum hits to inspect the complete hit list. Its pages display subsets; Export all hits (.txt) includes every recorded hit.
4. Export before starting another scan or reloading. The hit list is not a file that resumes the old scan.

The fixed QR settings are lowercase compressed-SEC hexadecimal, Byte mode, M error correction and version 5. The best natural mask pair in this lab is 0 ↔ 7. Its isolated-pair coverage is about 48.65% of the changed modules. “Optimum” describes that scoring rule, not proximity to a puzzle key.

Open 3D book plots recorded hits by page offset, line and mask direction. Select the bundled 138-hit scan, the current scan or an imported hit TXT. Drag to rotate, use the rotation/tilt controls, and select a dot or the hit navigation buttons.

Book side switches between recorded scalars and their mirrors. Previous opposite mask and Next opposite mask visit recorded hits with the other direction. Use hit as navigation anchor transfers the selection to exact book navigation. Lines connecting dots join the chosen records; omitted gaps have not been proven empty.

8. Anchors and experimental prediction

The bundled atlas starts with 519 independent anchor families and 1,038 points when their mirrors are included. It combines measured hits from multiple regions; it is a sparse atlas, not a complete map of the keyspace.

1. Enter a full SEC public key and range in Anchors. A mainnet P2PKH address instead enables exact comparison against recorded anchor addresses.
2. Set Candidate interval width and select Estimate page. An exact known match, an unverified estimate, insufficient references and address-only data are distinct outcomes.
3. Open proposed page inspects the proposed scalar. Prepare bounded search copies its point and interval into CPU Lab; it does not start searching. Pause an active CPU search before preparing a replacement.
4. Use Compare with controls to test unseen synthetic points at equal interval coverage. Export the test results to retain the evidence.

The predictor balances references across scalar regions, compares compressed-SEC bit differences and calculates a weighted scalar mean. Its rule has not established a reliable relation between public-key similarity and scalar distance. The bundled 256-target test at 10% coverage found 27 hits for anchor projection, 23 for uniform selection, 29 for the midpoint and 21 for random-anchor selection; those results did not establish an advantage.

Extend and reuse the atlas

In Sample distant regions, select the whole valid keyspace or the chosen prediction range, then set regions, scalars per region and a seed. The Anchors default is 16 × 2,048 = 32,768 checks, unlike Automatic’s 256 total default samples. Sampling discovers and records natural-mask anchors within those bounded windows.

Choose Pause task, then Export atlas + checkpoint to preserve the atlas and sampling positions. Import that JSON and choose Resume paused scan to continue. A comparison benchmark can be exported as evidence, but does not use the sampling resume control.

Add current optimum hits imports the current Pair Lab results. Automatic’s Add hits to Anchors transfers its recorded hits. Imported scalars, points and QR masks are verified before use. Export all hits .txt also works with the 3D book importer.

The distribution bars show record counts, not target probabilities. Clicking a region changes the selected bounds. A mirror is a derived reference point, not another independently scanned scalar range.

9. Pond and 4D Book

Pond: byte sensitivity and forecasts

First select a scalar in Keybook. In Pond, select one of its 32 bytes: byte 00 is most significant and byte 31 is least significant. Choose a low/high-bit flip or an addition/subtraction at that byte’s weight, then select Forecast → reveal. Cyan represents forecast movement and gold represents the measured movement.

Map all 32 bytes measures controlled disturbances across Scalar, WIF raw, SEC, X, HASH160, Address and QR. The seven-stage order is a comparison view; WIF and SEC are parallel representations of the scalar. An intervention that leaves the valid scalar range cannot be used.

4D Book: visualize the same measured disturbance

Choose the scalar byte and bit, then select Launch ripple. Adjust gravity, viscosity, drag, erosion, deposition or coupling to explore the visual simulation. Use Pause, Step, Reset and the time section to inspect its state.

Here X/Y are positions in a square field, Z is transformation depth, and T is simulation time. The forcing comes from measured transformation differences; the fluid and sediment are a visualization model. Energy, sediment and visual resemblance are not probabilities of recovering a key.

10. CPU Lab

1. Choose Target type or Auto detect, paste the target, and select the bounds notation.
2. Enter inclusive bounds and choose Exhaustive or Pollard kangaroo. Kangaroo requires a full SEC public key; use exhaustive comparison for an address or HASH160.
3. For HASH160 comparisons, choose compressed, uncompressed or both. For exhaustive order, choose forward or inward from both ends.
4. Set Work budget / run and select Start new. Watch the checked count, remaining interval and verification message.
5. Use Pause, then Export checkpoint before closing. Import that checkpoint in CPU Lab and select Resume to continue the saved session.

The optional kangaroo settings control walkers, collision-table size, pattern prepass and mirror collisions. An additional walker is a classical walk, not a guarantee of proportionally faster searching. Run safe benchmark measures a public test range; Simulate window builds a real classical oracle and plots ideal Grover probabilities.

Import targets or solved-puzzle rows

Import CSV / TSV / text reads addresses from the first column. Click a displayed P2PKH row to make it the active target. P2SH rows in this list are references only. Importing a list does not automatically launch a search against every address.

The separate solved-puzzle import requires these exact CSV headers:

bits,range_min,range_max,address,hash160_compressed,public_key,private_key

range_min, range_max and private_key are hexadecimal columns; bits is an integer. Each row’s public key, compressed hash and address are checked against its scalar. Predict page bands needs at least three suitable smaller-bit rows. Build range schedule creates metadata for intervals; it performs no candidate checks.

Automatic uses verified solved rows already loaded here when its sequence begins. The offline edition also retains its earlier prediction-comparison panel; its exported benchmark evidence is separate from the newer Anchors and Automatic reports.

11. Wormhole: coordinate index and signed search

Generate pages and look up saved coordinates

Enter a decimal page and choose Generate Page, or use Open Known Scalar. Generated records have exact scalar, coordinate, page and line information. Select a row to open it in the coordinate book, scalar page or Keybook.

Find in Saved Index accepts a full SEC key, X/Y, address or HASH160. It only examines records already saved by this browser. The coordinate index is not a precomputed index of every possible key. Its sorted index pages are different from the fixed scalar pages.

Digit constraints and page bridges

Suggest Exact Page finds the first value satisfying the 64-digit hex/wildcard mask. Its bounds are hexadecimal, and it also respects the current scalar-page floor. A mask-compatible value is not checked against an address by that action.

Calculate page bridge verifies the exact relation between two supplied page/line pairs; Reverse bridge swaps them. If both positions are known, their scalar difference is known. That identity does not provide an unknown target’s scalar.

Search a bounded public point

1. Paste the full target SEC, enter the inclusive interval and its notation, then set the table-entry cap and giant-lookup budget.
2. Select Plan from form to see the radius, stride and number of signed blocks.
3. Choose Start / Resume. A reported collision is accepted only after a fresh scalar multiplication reproduces the target point.
4. Use Pause and retain before Export checkpoint. Import the checkpoint and use Start / Resume to continue.

Changing a target or range requires a corresponding new plan. In the offline edition, reusable generator references can reduce setup work. Those cached references are not a record of target-search progress; preserve the search checkpoint separately.

12. Saving and resuming

CPU Lab and Wormhole also offer Restore tab autosave when a compatible session save exists. Keep explicit checkpoint files for work that matters; tab storage and browser storage can be cleared or unavailable.

The live site and offline address are different origins and have separate local data. Export and import the supported files when moving between them. A new software download includes the bundled samples and tools, not the work already held in your browser. Reports may include supplied scalars and WIF outputs; choose which files to share accordingly.

13. Website and offline editions

GoDaddy website: open the site's HTTPS address in a current desktop browser. No Python, PHP, database, or installation is required for visitors. Calculations and scans use your device; keep the tab open while they run.

Self-contained offline copy: download or copy the compiled index.html from the release package and open it in a current browser. Its scripts, styles, five worker engines, 3D Sphere, and manual are embedded. Browser storage and downloads can vary for local files.

Editable source edition: in the package's source folder, use Python 3 and run START_WINDOWS.bat, START_MAC_LINUX.command, or python3 start_offline.py. Open the loopback address printed by the launcher and keep it running. Stop it with Ctrl+C after exporting your work.

Export checkpoints before moving between the old site, this GoDaddy site, or a local file. Browser data belongs to the browser profile and origin; it is not copied by uploading the software. No saved personal sessions are included in this release.

14. Troubleshooting

15. Terms and equations

Scalar k

The integer used to generate a public curve point. Valid book scalars satisfy 1 ≤ k < n.

Public point Q

The point Q = kG. SEC is its compressed or uncompressed byte encoding.

HASH160 / address

A hash or encoded destination used for comparison. It is different from the full public point.

Mirror

Point negation: −Q has the same X and opposite Y modulo the field prime. Its scalar is n − k.

Natural optimum anchor

A recorded point whose compressed SEC and mirror receive natural QR masks 0 and 7 under the lab’s fixed settings.

Anchor family

A recorded scalar and its derived mirror treated as one independent reference family.

Budget

A limit on one method’s work. Its unit depends on the method: candidates, samples, trials, table entries, walk steps or lookups.

Checkpoint versus report

A checkpoint contains state supported by a particular resumable search. A report records what happened; Automatic report import restores setup only.

Exact page mapping

page(k) = 1 + floor((k − 1) / 1000)
line(k) = 1 + ((k − 1) mod 1000)
k = 1000 × (page − 1) + line

Exact anchor and mirror relations

A = aG
k = a + d                 Q = A + dG
n − k = (n − a) − d       −Q = −A − dG

These equations verify navigation when the scalar or offset is known. They do not determine the unknown offset from public-key similarity or a QR pattern. Experimental predictions remain separate until an exact target comparison succeeds.

End of manual · Visual Keybook, Automatic edition · 19 September 2026

16. 3D Sphere

The 3D Sphere tab embeds the full spherical viewer inside Visual Keybook. The original Keybook, Automatic, Anchors, Pond, 4D Book, CPU Lab, Wormhole, and their controls remain available. This Manual stays the final tab.

Explore and inspect

Drag to rotate, scroll or pinch to magnify. Click a star to inspect an exact inclusive interval. Double-click a region, or choose Enter region, to subdivide it. Parent moves up one level; Whole sphere returns to the entire valid scalar range. At 1,000 keys or fewer, each point is an individual scalar. Optical magnification does not change the represented interval.

Move between the views

Show current Keybook key sends the last successfully derived Keybook scalar to its exact page in the sphere. Follow key from whole sphere highlights its containing interval at the widest scale; enter the highlighted region repeatedly to reach the individual key. Open selection in Keybook rederives the selected individual scalar in the original QR/mirror laboratory. This button is disabled for ranges; it never substitutes a range midpoint or starts a search. The Keybook tab also has a View in 3D Sphere button, which first validates and derives its input.

Inspector, list, pins, and sessions

The sphere retains its own Inspector, Book, Pins, and Manual panels. Inspector includes decimal/hex/binary scalar, page and line, curve coordinates, public-key encodings, WIFs, address forms, scripts, and mirror partner. Book is an accessible clickable list of the visible ranges or keys. Pins mark reference keys. Save session / Load preserve the sphere view, camera, and pins in JSON; these do not replace Automatic, CPU, or Wormhole checkpoints. Private values are recalculated on import. No balance queries are made.

Display and performance

The viewer is initialized only when its tab first opens. Switching away pauses its animation and retains its view, key, and pins. Expand sphere fills the browser viewport; choose Exit expanded view to restore the tabs. On a narrow display the Inspector appears below the sphere and scrolls independently. The original animation controls and reduced-motion preference remain available.

Exactness and privacy

Large scalars pass between views as decimal strings and BigInt, never floating-point numbers. Page = (k−1)//1000+1 and line = (k−1)%1000+1. Star positions are a navigation layout, not a map of nearby addresses or a predictor of unknown keys. Neither opening this tab nor moving a key starts a scan. All bridge messages remain between this page and its embedded viewer. Do not enter funded wallet secrets; scalar/page/line and saved sessions are private-key material.