Release v0.3.0-alpha.20: fix histogram chart, polish layout

Polish pass on build_executive_report after live screenshot review
against NPV_of_a_capital_investment workbook.

Histogram chart was inverted (#18b)
Prior SetSourceData on the full 3-column [Bin, Count, Cumulative %]
block made Excel turn ALL THREE columns into data series — visible
result was bin centres plotted as bars (50k, 100k, 150k...), Count
plotted as a red line, Cumulative invisible. Fix: bind only to
Count + Cumulative columns, then explicitly assign
SeriesCollection(i).XValues to the Bin column on both series.
Now: proper frequency-histogram bars on bin centres, cumulative-%
line overlay on secondary Y axis.

Layout polish (column A now a narrow gutter)
- Content shifted right by one column: title band B:L (was A:J),
  headlines B-F (was A-E), output name at I (was G), stats table
  B-H (was A-G), callouts B-merged-to-L (was A-merged-to-J).
- Narrow gutters at A and M (width 2 each) for visual breathing room.
- Stats-table CV column widened to 16 — no more ##### overflow on
  values like 1.296.
- Alternating row tint on stats table for readability.
- High-CV cells bold in addition to coloured (survives print-to-PDF).
- Chart sizes bumped 400x240 / 360x240 (was 380x220 / 340x220).

399 tests pass. 4 existing tests updated for B-shifted columns.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: vosesoft

CHANGELOG.md

@@ -4,6 +4,30 @@ All notable changes to ModelRisk MCP. Follows [Keep a Changelog](https://keepach
 
 ## [Unreleased]
 
+## [0.3.0-alpha.20] — 2026-05-22
+
+Polish pass on `build_executive_report` after a live screenshot review against the `NPV_of_a_capital_investment` workbook. Two real issues found: the histogram chart was rendering with completely wrong semantics, and the layout was visually cramped with column A pulling double duty as label-holder and edge.
+
+### Fixed
+
+- **Histogram chart was inverted (#18b).** Prior versions called `SetSourceData` on the full 3-column block `[Bin, Count, Cumulative %]`, which made Excel turn all three columns into data series — visible result was "Bin" values plotted as random-height blue bars (the user's screenshot showed bars at 50k, 100k, 150k etc.), "Count" plotted as a red line, and Cumulative invisible. Fix: bind to ONLY the Count + Cumulative columns, then explicitly assign `SeriesCollection(i).XValues` on both series to the Bin column. Result: proper frequency-histogram bars with bin centres on the X axis, cumulative-% line overlay on a secondary Y axis. The chart now looks like what the report description says.
+
+### Polished
+
+- **Column A is now a narrow gutter (width 2).** Previously content started flush against the left edge with column A holding both labels AND being the page edge. Now the layout has narrow gutters at A and M, with content in B–L. Same change applied to the title band merge range (B:L instead of A:J), headline numbers (MEAN now at B6 instead of A6), stats table (Output at B26 instead of A26), and callouts (`•  ...` at B instead of A).
+- **Stats-table CV column no longer overflows.** Bumped width to 16 (was 14 implicit) so values like `1.296` render in full instead of `####`.
+- **Alternating row tint** on the stats table for readability when there are multiple outputs.
+- **High-CV cells now bold** in addition to coloured, so they survive print-to-PDF where colour fidelity drops.
+- **Chart sizes bumped** to 400×240 (histogram) and 360×240 (tornado) — the original 380×220 / 340×220 felt small relative to the title band. Charts shifted right by ~16pt to align with the new column-B content start.
+
+### Why this matters
+
+The report is the primary user-facing deliverable. A broken chart isn't "a bug to fix later" — it's the LLM lying to the user about what got built (`chart_count: 2` while the chart was visually wrong). Same goes for the cramped layout: a stakeholder sees the report and forms an opinion about modelrisk-mcp from that single screenshot. Worth getting right.
+
+### Tests
+
+399 unit tests pass. 4 existing tests updated to assert against the new column positions (B-shifted).
+
 ## [0.3.0-alpha.19] — 2026-05-22
 
 Fixes the bug-#23 lookup-after-samples regression discovered while end-to-end testing alpha.18 against a real workbook: `get_sensitivity_ranking` returned empty on the first call after `run_simulation`, then worked on the second identical call. The diagnostic trace was unambiguous — the output looked up fine, its samples loaded, then every input lookup against the same handle returned None.

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "modelrisk-mcp"
-version = "0.3.0a19"
+version = "0.3.0a20"
 description = "Open MCP server bridging Anthropic Claude (and any MCP-compatible client) with the ModelRisk Excel add-in."
 readme = "README.md"
 requires-python = ">=3.11"

server.json

@@ -3,12 +3,12 @@
   "name": "io.github.vosesoftware/modelrisk-mcp",
   "title": "ModelRisk",
   "description": "Read, build, fit, and run Monte Carlo risk models in Excel through Vose Software's ModelRisk.",
-  "version": "0.3.0a19",
+  "version": "0.3.0a20",
   "packages": [
     {
       "registryType": "pypi",
       "identifier": "modelrisk-mcp",
-      "version": "0.3.0a19",
+      "version": "0.3.0a20",
       "transport": {
         "type": "stdio"
       }

src/modelrisk_mcp/__init__.py

@@ -1 +1 @@
-__version__ = "0.3.0a19"
+__version__ = "0.3.0a20"

src/modelrisk_mcp/bridge/reports.py

@@ -267,8 +267,9 @@ def test_creates_sheet_with_title_and_subtitle(self) -> None:
         )
         assert isinstance(result, ExecutiveReportResult)
         sheet = book.sheets["Report"]
-        assert sheet.cells.get("A1") == "NPV under uncertainty"
-        assert sheet.cells.get("A2") == "5,000 iterations · seed 1 · 2026-05-22"
+        # alpha.20 layout shift: content starts at column B (A is gutter).
+        assert sheet.cells.get("B1") == "NPV under uncertainty"
+        assert sheet.cells.get("B2") == "5,000 iterations · seed 1 · 2026-05-22"
         assert sheet.activated is True
 
     def test_headline_numbers_in_expected_cells(self) -> None:
@@ -284,12 +285,13 @@ def test_headline_numbers_in_expected_cells(self) -> None:
             sensitivity=_make_sensitivity(),
         )
         sheet = book.sheets["Report"]
-        assert sheet.cells.get("A6") == 1000.0  # Mean
-        assert sheet.cells.get("C6") == 1000.0  # P50 = mean for symmetric
-        assert sheet.cells.get("E6") == 200.0   # Stdev
+        # alpha.20 layout: B=Mean, C=P5, D=P50, E=P-hi, F=Stdev
+        assert sheet.cells.get("B6") == 1000.0  # Mean
+        assert sheet.cells.get("D6") == 1000.0  # P50 = mean for symmetric
+        assert sheet.cells.get("F6") == 200.0   # Stdev
         # Headline label row
-        assert sheet.cells.get("A5") == "MEAN"
-        assert sheet.cells.get("E5") == "STDEV"
+        assert sheet.cells.get("B5") == "MEAN"
+        assert sheet.cells.get("F5") == "STDEV"
 
     def test_stats_table_includes_secondary_outputs(self) -> None:
         book = _FakeBook()
@@ -309,11 +311,11 @@ def test_stats_table_includes_secondary_outputs(self) -> None:
             secondary_results=secondary,
         )
         sheet = book.sheets["Report"]
-        # Stats table starts at row 26 (header), data at 27+
-        assert sheet.cells.get("A26") == "Output"
-        assert sheet.cells.get("A27") == "Profit"      # primary first
-        assert sheet.cells.get("A28") == "Cost"
-        assert sheet.cells.get("A29") == "Revenue"
+        # alpha.20 layout: stats table shifted to columns B..H.
+        assert sheet.cells.get("B26") == "Output"
+        assert sheet.cells.get("B27") == "Profit"      # primary first
+        assert sheet.cells.get("B28") == "Cost"
+        assert sheet.cells.get("B29") == "Revenue"
 
     def test_callouts_generated_from_data(self) -> None:
         book = _FakeBook()
@@ -334,7 +336,7 @@ def test_callouts_generated_from_data(self) -> None:
         # CALLOUTS header on first callout row.
         callout_cells = [
             v for k, v in sheet.cells.items()
-            if k.startswith("A") and isinstance(v, str)
+            if k.startswith("B") and isinstance(v, str)
             and v.startswith("•")
         ]
         assert len(callout_cells) >= 2
@@ -363,7 +365,8 @@ def test_replaces_existing_sheet(self) -> None:
         # New sheet has the fresh content.
         new_sheet = book.sheets["Report"]
         assert new_sheet is not existing
-        assert new_sheet.cells.get("A1") == "fresh"
+        # alpha.20: title shifted from A1 to B1.
+        assert new_sheet.cells.get("B1") == "fresh"
 
     def test_empty_samples_doesnt_crash(self) -> None:
         """If get_samples returned [] (no .vmrs / empty result), the

tests/unit/test_reports.py

@@ -5,7 +5,7 @@
 
 
 def test_version_is_set() -> None:
-    assert __version__ == "0.3.0a19"
+    assert __version__ == "0.3.0a20"
 
 
 def test_server_name() -> None: