TrustLens
A serious analysis workbench for one formula. TrustLens puts a flow diagram, a plain-English explanation, a chain-of-trust verdict, the formula's references, and an optional editor in a single floating window — so you can understand a number, judge whether it can be trusted, and fix it in place.
What it is
TrustLens opens a floating, resizable window anchored to Excel titled Modelxcel TrustLens. It is a multi-region workbench: a header with the cell address and a source-mode badge (e.g. “Scalar · A1”, “Dynamic Anchor · R1C1”); an optional Formula Editor band; a proof strip showing a color-coded verdict; a main split with the Flow Diagram on the left (~70%) and a tabbed details panel on the right (~30%); and a footer with a zoom slider and Copy / Save buttons. The three detail tabs are Explain, Proof, and References.
Flow Diagram (left panel)
The selected cell is the root node; its precedents form a navigable graph. Click a node to drive the detail tabs and proof strip for that node. Navigate with a breadcrumb (Alt+Left/Alt+Right back/forward, Ctrl+Home to root); a preview drawer previews a node before you commit to it. Ctrl+Shift+F fits the diagram; Ctrl+Shift+P prunes to the active path. Copy puts the diagram image on the clipboard; Save exports PNG/JPG/BMP. External-workbook references appear in the References tab but are not rendered as diagram nodes.
Explore a formula without losing your place
TrustLens is built for non-destructive exploration: you can follow a number down through several precedents and always retrace your way back.
The preview drawer
Click a node in the flow diagram and a preview drawer opens at the bottom of the diagram without changing what TrustLens is focused on. The drawer shows the target’s formula, a one-line summary, and a small diagram of its precedents — so you can size up a cell before diving in. The action button adapts to where the target sits in your path:
- Focus Here — when the target is new. It commits the preview as the new root and re-centers TrustLens (and the Explain / Proof / References tabs) on that cell.
- Jump to Step — when the target is already in your breadcrumb path. Rather than add a duplicate, it jumps back to that existing step.
Press Esc to dismiss the drawer without moving. Press Enter to commit an open preview (Focus Here / Jump to Step); Alt+Up commits only when the preview target is already in your breadcrumb path (the Jump to Step case).
The breadcrumb trail
Every Focus Here adds a step to a breadcrumb across the top of the diagram. Click any earlier step to jump straight back to it; use Alt+Left / Alt+Right to move back and forward and Ctrl+Home to return to the original cell (the root). TrustLens captures each step’s diagram zoom and position, so jumping back lands you exactly where you were.
Explain view
For the selected node, a summary card (with a severity-colored accent bar — neutral, warning, or suspicious) gives the sub-expression label, a plain-English explanation, and the current value/address; below it an evidence list of up to 80 rows surfaces diagnostics (lookup behavior, aggregation, findings) with a label, detail, and source address. It is read-only. “No explanatory evidence for this selection.” is normal for constants or pure references.
Proof view
The chain-of-trust for the formula’s precedent chain, drawn from the same ledger as Proof & Trust. A verdict card reads Chain intact, Review needed (drift detected), Chain broken (a hash link failed), No provenance, or Proof unavailable/queued. Six metric tiles count Bound, Signed, Unattested, Drifted, Records, and Cells (chain nodes). A timeline lists each chain node with a badge (BOUND / SIGNED / DRIFT / VECTOR / BLOCK / NO RECORD …); selecting a row shows its full ledger facts (binding kind, signer, vector/block scope, latest event, drift). The chain is capped at 80 nodes and is single-workbook. (When the inline editor has a staged edit, a Pending card previews the proof record that would be appended on commit.)
References view
Every reference parsed from the formula, summarized as three tiles (Same sheet / Off sheet / Named) and listed with category chips. Click a row to highlight it; double-click or Enter to jump Excel’s selection to that cell/range. With the editor open, moving the caret over a reference token highlights the matching row here and the matching node in the diagram (see Trace mode).
Trace mode & cross-section sync
Toggle Trace mode in the editor (Ctrl+T). As your caret moves over a reference token, TrustLens highlights the same reference across the diagram node and the references list (always-on Level-1 sync), and — in Trace mode — can also follow the selection in Excel(Level-2). F12 or Ctrl+Click on a token does a one-shot jump without toggling Trace.
Edit a formula in place
TrustLens embeds a real formula editor so you can fix what you find without leaving the analysis. It has two modes: Compact (read-only and low-distraction — the formula is shown for review) and Expanded (the full editing surface). Press Alt+E to switch. Nothing is written until you explicitly commit.
In Expanded mode a toolbar offers Commit, Revert, Beautify, Minify, Preview, and Trace — each also available by keyboard:
| Key | What it does |
|---|---|
Ctrl+Space | Autocomplete over Excel functions and the workbook’s defined names (Expanded). |
F4 | Cycle the A1 reference anchor under the caret: $A$1 → A$1 → $A1 → A1 (Expanded, A1 mode). |
Ctrl+I / Ctrl+M | Beautify (indent, multi-line) or Minify (single line) the staged text — editor-only, no cell write. |
F5 | Preview: evaluate your staged edit without committing — it temporarily writes the staged formula, reads the result, then restores the cell. |
Ctrl+T | Toggle Trace mode — caret-over-reference highlights sync to the diagram and references list, and Excel’s selection follows your caret. |
F12 / Ctrl+Click | One-shot jump: move Excel’s selection to the reference under the caret (no Trace toggle needed). |
Ctrl+Enter / Esc | Commit the edit, or revert (discard the staged change). |
Editing is explicit, and the commit clears Excel's undo
Nothing is written until you press Ctrl+Enter. That first writes the staged formula to the cell, then attempts to append the audit/proof record to the ledger. The cell write clears Excel’s native undo stack — once committed, Ctrl+Z cannot revert it. Before committing you can revert with Esc.
The two steps are not guaranteed to both succeed. If the cell write succeeds but the audit append fails, TrustLens leaves the edit marked dirty, shows a warning, and lets you retry the ledger append (a repeat Ctrl+Enter on the same cell and formula) — the cell already holds the new formula. If you edit cell A then click cell B while the edit is dirty, TrustLens preserves A’s staged text and reminds you the commit targets the original cell.
Aside from an explicit commit (and the transient F5 preview write-and-restore), TrustLens writes nothing to the workbook — browsing views, navigating, and previewing change no data; Copy/Save only export the diagram image.
A worked example
Say an EBITDA cell looks wrong. Open TrustLens on it (Ctrl+Shift+I) and click a precedent node — say the revenue driver. The preview drawer opens so you can read its formula and summary without moving; click Focus Here to make it the new root. Check its Proof tab (is the input bound, the formula signed?) and its References. Turn on Trace mode (Ctrl+T) and run your caret across the formula to watch each input light up on the grid. Find a wrong reference? Switch to Expanded, fix it, press F5 to preview the corrected result, and press Ctrl+Enter only when you’re satisfied — then use the breadcrumb to walk back up to EBITDA and confirm the change flowed through. Throughout, you never lost your place.
Keyboard shortcuts
| Key | Action |
|---|---|
Ctrl+Shift+I | Open TrustLens for the active cell |
Ctrl+Tab / Ctrl+Shift+Tab | Cycle Explain → Proof → References |
Alt+Left / Alt+Right / Ctrl+Home | Diagram back / forward / root |
Ctrl+Shift+F / Ctrl+Shift+P | Fit diagram / toggle prune |
Alt+E | Toggle the Formula Editor (Compact / Expanded) |
Alt+P | Pick a reference by selecting it in Excel |
Ctrl+T · F12 · Ctrl+Click | Trace mode · one-shot jump to reference under caret |
Ctrl+Enter · Esc | Commit edit (writes + clears undo) · revert |
? / F1 | Show the in-window shortcut help |
How TrustLens connects to Proof & Trust
The Proof view is built from the workbook’s causality ledger and registry — the same store documented under Proof & Trust. The most complete chain comes from the “Why This Number” causal walker; the verdict, badges, and per-node facts mirror what Why This Number reports. When you commit a formula edit, the proof record it appends shows up in this same ledger. TrustLens is where Formula Tools and the Proof system meet: understand the formula, see whether its inputs are bound and its logic signed, and edit with an auditable trail.
Scope & troubleshooting
The proof chain and diagram are single-workbook; cross-workbook reference identities can be listed in References but are not rendered as diagram nodes. If the editor band is missing, the Scintilla editor failed to load — the rest of the workbench still works. “Chain broken” means a ledger hash link is invalid; “Review needed” means a cell in the chain has drifted (recorded formula ≠ live formula); “No provenance” means nothing in the chain has been bound or signed yet.