fix(UtilitiesCS.Test): remove duplicate PercentageFormatterTests compile entry (CS2002)

.claude/agent-memory/feature-review/project_durable-feature-script-triggers-python-coverage-gate.md

@@ -10,3 +10,5 @@ Issue #354 (stale-app-config-binding-redirects): the atomic-planner's `durable-s
 **Why:** The mandatory coverage rule in the feature-review SKILL has no carve-out for "tooling scripts committed to a feature folder rather than production `src/`." Treating it as exempt would be an unauthorized scope-narrowing.
 
 **How to apply:** When a feature-folder diff includes a new `.py`/`.ps1`/`.ts`/`.cs` file under `scripts/` (not just under conventional production paths), still classify it as a changed-file language and require the coverage row. Independently verify the script's *correctness* (re-run it, diff its output, run Black/Ruff/Pyright directly) as strong corroborating evidence for the code-review, but do not let that substitute for the missing coverage artifact in the policy-audit verdict. See [[project_coverage-hook-label-substring-false-positive]] for the adjacent Test-LanguageCoverageRow line-matching mechanics used to write a compliant FAIL row.
+
+**Confirmed instance under `evidence/` (not just `scripts/`)**: #394 (utilitiescs-test-cs2002-duplicate-compile-entry) — an executor retained a `.ps1` sweep helper under `<FEATURE>/evidence/baseline/duplicate-sweep.ps1` "for reproducibility" even though the plan only required capturing the script's *output* in a `.md` artifact. `Get-ChangedLanguageSet` matches any `.ps1`/`.py` bullet regardless of path, so the same mandatory-coverage-gate logic applies to committed evidence-tree scripts, not just `<FEATURE>/scripts/`. Recommended remediation in that case: delete the retained script (its logic/output is already durably captured in the sibling timestamped `.md` narrative) rather than build out Pester/PoshQC/coverage tooling for a one-off audit helper — this is the proportionate fix for a scope-locked bugfix and avoids the unfunded toolchain obligation entirely.

UtilitiesCS.Test/UtilitiesCS.Test.csproj

@@ -353,7 +353,6 @@
     <Compile Include="OutlookObjects\Folder\FolderScorerRegressionTests.cs" />
     <Compile Include="OutlookObjects\Folder\FolderScoreTests.cs" />
     <Compile Include="OutlookObjects\Folder\FolderScorerTests.cs" />
-    <Compile Include="OutlookObjects\Folder\PercentageFormatterTests.cs" />
     <Compile Include="OutlookObjects\Folder\FolderNodeViewModelTests.cs" />
     <Compile Include="OutlookObjects\Folder\FolderHierarchyBuilderTests.cs" />
     <Compile Include="OutlookObjects\Folder\FolderTreeStateModelTests.cs" />

docs/features/active/2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394/code-review.2026-08-10T23-45.md

@@ -0,0 +1,28 @@
+# Code Review — utilitiescs-test-cs2002-duplicate-compile-entry-394
+
+- **Issue:** #394
+- **Branch:** `bug/utilitiescs-test-cs2002-duplicate-compile-entry-394`
+- **Base:** `origin/epic/build-ci-coverage-gate-fidelity-integration` (merge-base `a5e336e5`)
+- **Timestamp:** 2026-08-10T23-45
+
+## Executive Summary
+
+The substantive code change — deletion of one duplicate `<Compile Include>` item from `UtilitiesCS.Test/UtilitiesCS.Test.csproj` — is minimal, correctly targeted, and free of quality issues. No production `.cs` code, class, or method is added or modified. Review findings below are limited to (a) the newly committed evidence-tree helper script, which does not meet the repository's PowerShell coding/toolchain standards, and (b) minor documentation-accuracy discrepancies in `spec.md`'s duplicate-sweep table versus the actual captured sweep evidence. Neither the deleted line nor the surrounding `.csproj` content shows any stylistic or correctness issue.
+
+## Findings Table
+
+| Severity | File | Location | Finding | Recommendation | Rationale | Evidence |
+|---|---|---|---|---|---|---|
+| Blocking | `docs/.../evidence/baseline/duplicate-sweep.ps1` | whole file | Committed PowerShell file with no PoshQC format/analyze run, no Pester test, and no coverage artifact; also cross-references the mandatory Coverage Verification policy finding recorded in `policy-audit.2026-08-10T23-45.md` | Remove the script from the committed tree (its logic and full output are already durably captured in `evidence/baseline/duplicate-sweep.2026-08-10T22-31.md`); if retained, run PoshQC format/analyze, add a Pester test, and produce `artifacts/pester/powershell-coverage.xml` | `.claude/rules/powershell.md` mandates format -> analyze -> test for every `**/*.ps1` file with no evidence-tree carve-out; `general-unit-test.md` requires >=85%/>=75% coverage per language with changed files | `git diff --numstat` shows the file as newly added (27 insertions); `artifacts/pester/` does not exist in the worktree |
+| Minor | `docs/.../evidence/baseline/duplicate-sweep.ps1` | whole file (lines 1-27) | Script is a bare top-level script with no `[CmdletBinding()]`, no advanced function, and no parameter validation; also has no `try`/`catch` around `Get-Content -Raw`/`[xml]` casts, so a missing or malformed input file throws an unhandled terminating error rather than a clear diagnostic | Wrap the sweep logic in an advanced function (or, per the Blocking finding above, remove the script entirely) | `.claude/rules/powershell.md` "Coding Standards": prefer advanced functions with `CmdletBinding()`; general-code-change.md "Fail fast and explicitly" | Direct read of the script contents |
+| Minor | `docs/.../spec.md` | `## Root Cause Analysis` -> "Duplicate Sweep Result" table | Table states `Analyzer` count = 9, `Reference` count ~= 114, and `packages.config` count ~= 99; the actual captured sweep evidence (`evidence/baseline/duplicate-sweep.2026-08-10T22-31.md`) reports `Analyzer Total=11`, `Reference Total=126`, `packages.config Total=105` | Update the table's precise `Analyzer` figure (stated without an approximation qualifier, unlike `Reference`/`packages.config`) to match the captured evidence, or add a `~` qualifier consistently | Does not change the duplicate-finding conclusion (zero duplicates in either count for these item types), but a precise, unqualified count that does not match its own cited evidence is a documentation-accuracy defect | Side-by-side comparison of `spec.md` lines 158-169 against `evidence/baseline/duplicate-sweep.2026-08-10T22-31.md` |
+| Informational | `UtilitiesCS.Test/UtilitiesCS.Test.csproj` | line 356 (removed) | The deletion is exactly one line, correctly identifies the second (redundant) of the two identical `<Compile Include>` items, and leaves the first occurrence (line 304) and all surrounding lines untouched | None — this is the correct fix | Matches spec.md's Proposed Fix design exactly; confirmed no reordering/line-ending churn via `git diff` | `git diff origin/epic/build-ci-coverage-gate-fidelity-integration...HEAD -- UtilitiesCS.Test/UtilitiesCS.Test.csproj` |
+| Informational | Evidence artifacts (all) | `evidence/baseline/`, `evidence/qa-gates/`, `evidence/regression-testing/` | All command-bearing artifacts carry `Timestamp:`, `Command:`, `EXIT_CODE:`, and `Output Summary:` fields; no `EXIT_CODE: SKIPPED` occurrences; fail-before and post-fix builds use the identical `/t:Rebuild` command | None | Meets `evidence-and-timestamp-conventions` schema requirements | Direct read of each artifact |
+
+## Design and API Impact
+
+Not applicable — no class, method, interface, or public API is added, removed, or changed. This is a build-configuration item-list edit.
+
+## Overall Assessment
+
+The bugfix itself is correctly scoped, minimal, and well-evidenced. The one blocking finding is a policy/toolchain-compliance issue caused by an unplanned decision to commit a helper script into the evidence tree, not a defect in the production fix. See `policy-audit.2026-08-10T23-45.md` Section 4 and `remediation-inputs.2026-08-10T23-45.md` for the required remediation.

docs/features/active/2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394/code-review.2026-08-11T04-05.md

@@ -0,0 +1,30 @@
+# Code Review — utilitiescs-test-cs2002-duplicate-compile-entry-394 (Reaudit — Remediation Cycle 1 Exit Gate)
+
+- **Issue:** #394
+- **Branch:** `bug/utilitiescs-test-cs2002-duplicate-compile-entry-394` (HEAD `f39c6fc9`)
+- **Base:** `origin/epic/build-ci-coverage-gate-fidelity-integration`
+- **Timestamp:** 2026-08-11T04-05
+- **Entry review (unmodified, retained):** `code-review.2026-08-10T23-45.md`
+
+## Executive Summary
+
+This reaudit evaluates only the delta introduced by remediation cycle 1 (commits `2a2116eb`, `f39c6fc9`), since the entry review already found the underlying CS2002 fix itself free of quality issues (confirmed unregressed this cycle — see policy-audit reaudit Section 2). The remediation cycle: (1) removes the previously-flagged committed PowerShell helper script, resolving both blocking findings the entry review raised against it; (2) corrects three stale numeric figures in `spec.md`'s duplicate-sweep table; (3) adds an `analyzer-not-applicable` evidence artifact for parity with sibling determinations; and (4) revises one remediation-plan task's acceptance wording mid-execution to accommodate a legitimate subagent memory-index byproduct. No new quality defect is introduced.
+
+## Findings Table
+
+| Severity | File | Location | Finding | Recommendation | Rationale | Evidence |
+|---|---|---|---|---|---|---|
+| Informational | `docs/.../evidence/baseline/duplicate-sweep.ps1` | (deleted) | The entry review's two findings against this file (Blocking: no coverage artifact/toolchain run; Minor: no `[CmdletBinding()]`, no error handling) are both resolved by deletion rather than remediation of the script itself | None — deletion is the correct proportionate response for a one-off audit helper whose logic and full output are already durably captured verbatim in `evidence/baseline/duplicate-sweep.2026-08-10T22-31.md` | `general-code-change.md`'s "temporary throwaway scripts" allowance; avoids an unfunded Pester/PoshQC/coverage-tooling obligation for a script with no lasting reuse value | `git diff --name-only` shows zero `.ps1` files in the current diff; file confirmed absent from disk |
+| Resolved | `docs/.../spec.md` | `## Root Cause Analysis` -> "Duplicate Sweep Result" table | The entry review's Minor finding (Analyzer=9 vs actual 11, Reference=~114 vs actual 126, packages.config=~99 vs actual 105) is corrected | None — figures now read `Analyzer`=11, `Reference`=126, `packages.config`=105, matching `evidence/baseline/duplicate-sweep.2026-08-10T22-31.md` exactly | Documentation accuracy; independently re-verified numerically in this reaudit's policy-audit Section 5 | `evidence/qa-gates/post-remediation-spec-table.2026-08-10T23-45.md`; direct read of current `spec.md` lines 160-169 |
+| Resolved | `docs/.../evidence/qa-gates/analyzer-not-applicable.2026-08-10T23-45.md` | new file | The feature-audit's non-blocking observation (no dedicated analyzer-not-applicable artifact for evidentiary parity with the CSharpier/nullable determinations) is addressed | None — artifact added, states the zero-`.cs`-files rationale, cross-references the two sibling determinations | Evidentiary parity | Direct read of the artifact |
+| Informational | `remediation-plan.2026-08-10T23-45.md` | `[P2-T2]` | Acceptance clause revised mid-execution from a blanket "no path outside `<FEATURE>/` other than the csproj" to four explicit clauses (a)-(d), carving out `.claude/agent-memory/**` while re-affirming (and making explicit) the `CLAUDE.md`/`.claude/rules/`/`scripts/`/`.cs`/other-`.csproj`/coverage-threshold exclusions and, critically, stating the `.ps1`/`.psm1`/`.psd1` check is exempt from no carve-out | None — the revision is judged legitimate; it does not weaken any check this remediation cycle exists to enforce | Trigger was a genuine, verified two-line append to a pre-existing feature-review subagent memory file — a delegation byproduct, not a scope violation | Diff of `2a2116eb` vs `f39c6fc9` versions of the plan file; `git diff` of the memory-index file itself (2 insertions, 0 deletions) — see policy-audit reaudit Section 3 |
+| Informational | `UtilitiesCS.Test/UtilitiesCS.Test.csproj` | (unchanged this cycle) | No line of this file is touched by the remediation cycle; the single-line deletion from the original fix (`f58f8474`) remains intact and unregressed | None | Confirms the remediation cycle did not reopen or perturb the underlying fix | `git diff origin/epic/build-ci-coverage-gate-fidelity-integration...HEAD -- UtilitiesCS.Test/UtilitiesCS.Test.csproj` shows `1 -`, `0 +` |
+| Informational | Remediation-cycle evidence artifacts (all) | `evidence/other/`, `evidence/qa-gates/`, `evidence/remediation-baseline/` | All 15 remediation-plan tasks' artifacts carry `Timestamp:`, `Command:`, `EXIT_CODE:`, and `Output Summary:` fields; no `EXIT_CODE: SKIPPED` occurrences; `[P2-T2]`'s evidence file explicitly documents its own superseding re-run after the mid-execution revision rather than silently replacing the earlier capture | None | Meets `evidence-and-timestamp-conventions` schema requirements; the explicit "supersedes" framing is good practice for auditability | Direct read of each artifact |
+
+## Design and API Impact
+
+Not applicable — no class, method, interface, or public API is added, removed, or changed by this remediation cycle. The cycle is documentation/evidence-tree maintenance plus one file deletion (a non-production PowerShell helper).
+
+## Overall Assessment
+
+The remediation cycle correctly and proportionately closes the entry review's blocking finding by removing the unplanned PowerShell helper script rather than building out unfunded toolchain/coverage infrastructure for it. The three stale-figure corrections in `spec.md` are numerically accurate. The one mid-execution acceptance-wording revision is judged legitimate on inspection of both the original and revised text: it narrows the exclusion to exactly the byproduct that triggered it while explicitly re-stating (and strengthening, via explicit non-exemption) the substantive checks the original clause protected. No new code-quality or policy-compliance defect is introduced by this cycle.

docs/features/active/2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394/evidence/baseline/baseline-test-count.2026-08-10T22-31.md

@@ -0,0 +1,22 @@
+Timestamp: 2026-08-10T22-31
+
+Command: `pwsh -NoProfile -Command "& 'C:\Program Files\Microsoft Visual Studio\18\Community\Common7\IDE\Extensions\TestPlatform\vstest.console.exe' UtilitiesCS.Test\bin\Debug\UtilitiesCS.Test.dll /InIsolation '/TestCaseFilter:FullyQualifiedName~PercentageFormatterTests'"`
+
+EXIT_CODE: 0
+
+Output Summary:
+```
+Passed FormatPercent_Zero_ReturnsZeroPercent [46 ms]
+Passed FormatPercent_One_ReturnsHundredPercent [< 1 ms]
+Passed FormatPercent_TypicalValue_RoundsToWholePercent [< 1 ms]
+Passed FormatPercent_RoundsDownBelowMidpoint [< 1 ms]
+Passed FormatPercent_AtMidpoint_RoundsAwayFromZero [< 1 ms]
+Passed FormatPercent_SmallMidpoint_RoundsAwayFromZero [< 1 ms]
+Passed FormatPercent_Null_ReturnsEmptyString [< 1 ms]
+
+Test Run Successful.
+Total tests: 7
+     Passed: 7
+```
+
+Baseline scoped test count against the (still-duplicated) pre-fix rebuilt assembly (`UtilitiesCS.Test\bin\Debug\UtilitiesCS.Test.dll`, produced by the P0-T9 `/t:Rebuild`) is exactly 7 tests, all passed, matching the spec's documented count of 7 `[TestMethod]` members in `PercentageFormatterTests.cs`. This confirms the duplicate `<Compile>` item does not change the number of discoverable tests (it only causes the file to be passed to `csc.exe` twice).

docs/features/active/2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394/evidence/baseline/duplicate-sweep.2026-08-10T22-31.md

@@ -0,0 +1,27 @@
+Timestamp: 2026-08-10T22-31
+
+Command: `pwsh -NoProfile -File "docs\features\active\2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394\evidence\baseline\duplicate-sweep.ps1"`
+
+The sweep script (`duplicate-sweep.ps1`, stored alongside this artifact for reproducibility) XML-parses `UtilitiesCS.Test\UtilitiesCS.Test.csproj` and groups `Include` attribute values by item type (`Compile`, `EmbeddedResource`, `None`, `Reference`, `ProjectReference`, `BootstrapperPackage`, `Analyzer`, `AdditionalFiles`), and separately parses `UtilitiesCS.Test\packages.config` and groups `<package>` `id` attribute values.
+
+EXIT_CODE: 0
+
+Raw output:
+```
+ItemType=Compile Total=452 DuplicateIncludeValues=2
+  DUPLICATE: 'OutlookObjects\Folder\PercentageFormatterTests.cs' x2
+ItemType=EmbeddedResource Total=1 DuplicateIncludeValues=0
+ItemType=None Total=7 DuplicateIncludeValues=0
+ItemType=Reference Total=126 DuplicateIncludeValues=0
+ItemType=ProjectReference Total=2 DuplicateIncludeValues=0
+ItemType=BootstrapperPackage Total=2 DuplicateIncludeValues=0
+ItemType=Analyzer Total=11 DuplicateIncludeValues=0
+ItemType=AdditionalFiles Total=1 DuplicateIncludeValues=0
+packages.config Total=105 DuplicateIds=0
+```
+
+Output Summary: The sweep confirms exactly one duplicate `Include` value in the entire project file: `Compile` / `OutlookObjects\Folder\PercentageFormatterTests.cs`, appearing twice (lines 304 and 356, per P0-T7). Zero duplicates were found in every other item type (`EmbeddedResource`, `None`, `Reference`, `ProjectReference`, `BootstrapperPackage`, `Analyzer`, `AdditionalFiles`) and zero duplicate `id` values in `packages.config`.
+
+Note on the `DuplicateIncludeValues=2` label for `Compile`: this is a PowerShell array-unwrapping artifact, not a second duplicate group. When `Where-Object` returns exactly one matching `Group-Object` result, PowerShell unwraps the single-element array to a scalar `GroupInfo` object, so `.Count` on the (would-be) array instead returns that group's own `.Count` property — the occurrence count of the one duplicate value (2), not the number of distinct duplicate values (1). The enumerated `DUPLICATE:` lines are the authoritative per-group finding and confirm exactly one duplicate `Include` value in `Compile` and none elsewhere; `Total=452` for `Compile` matches the plan's stated baseline `<Compile Include=` count. Verified independently: for item types with zero duplicates, PowerShell's `$null.Count` evaluates to `0`, which is why those lines correctly read `DuplicateIncludeValues=0`.
+
+This confirms the plan's acceptance criterion for P0-T8: exactly one duplicate (`Compile` / `PercentageFormatterTests.cs`) and zero duplicates in every other item type and in `packages.config`.

docs/features/active/2026-08-10-utilitiescs-test-cs2002-duplicate-compile-entry-394/evidence/baseline/fail-before-cs2002.2026-08-10T22-31.md

@@ -0,0 +1,15 @@
+Timestamp: 2026-08-10T22-31
+
+Command: `pwsh -NoProfile -Command "& 'C:\Program Files\Microsoft Visual Studio\18\Community\MSBuild\Current\Bin\MSBuild.exe' UtilitiesCS.Test\UtilitiesCS.Test.csproj /t:Rebuild /p:Configuration=Debug /p:Platform=AnyCPU"`
+
+EXIT_CODE: 0 (Build succeeded — CS2002 is a warning, not a build failure)
+
+Output Summary: `/t:Rebuild` forced a genuine `CoreCompile` for `UtilitiesCS.Test.csproj` and its project-reference dependency chain (`TaskMaster`, `QuickFiler`, `TaskVisualization`, `Tags`, `UtilitiesCS`, `ToDoModel`). Build output contains the literal warning:
+
+```
+CSC : warning CS2002: Source file 'C:\Users\DanMoisan\repos\TaskMaster\.claude\worktrees\agent-a267ee5c24c8a630d\UtilitiesCS.Test\OutlookObjects\Folder\PercentageFormatterTests.cs' specified multiple times [C:\Users\DanMoisan\repos\TaskMaster\.claude\worktrees\agent-a267ee5c24c8a630d\UtilitiesCS.Test\UtilitiesCS.Test.csproj]
+```
+
+Final summary line: "6 Warning(s), 0 Error(s)". Total of 12 lines in the captured output contain the word "warning" (5 duplicate `System.Reactive.PackagesConfigCheck.targets` packages.config-migration warnings emitted once per dependent project in the chain, plus their MSBuild-log restatement lines, plus the one CS2002 line and its restatement). The CS2002 substring appears exactly 2 times in the raw output (once under the `CoreCompile` target block, once implicitly reiterated as part of the same block's full-path text match) — confirming the pre-change baseline reproduces the CS2002 warning for `PercentageFormatterTests.cs` as required. This satisfies the fail-before evidence requirement: a `/t:Build` capture would not be acceptable per spec.md; this is a genuine `/t:Rebuild`.
+
+Full raw MSBuild output is not persisted verbatim in this artifact (large, includes hundreds of resource-DLL copy lines from the Rebuild of dependency projects); the CS2002 warning line and the build summary line above are the load-bearing excerpts and were verified directly against the captured command output.