feat: verify S235 @demithras bounty — NO verdict, duplicate of S034 (#542)

[xliry]

Issue: #204
Submission: #204 (comment)
Author: @demithras

Problem (in our own words)

The submission claims that 87 import statements across 55 files in app/ and intelligence/ bypass engine's underscore-prefixed private subpackages (_state, _scoring, _plan, _work_queue), violating the encapsulation boundary signaled by the naming convention. It also claims a state.py facade exists that re-exports selected symbols.

Evidence

Counts verified at commit 6eb2065:

• app/: 57 private imports (matches S034 exactly)
• intelligence/: 30 private imports
• Total: 87 imports across 55 files (both numbers exact)
• Breakdown: _state:32, _work_queue:24, _scoring:24, _plan:7
• StateModel import count: 27 (submission claims 24)
• Facade: only plan.py exists (241 lines, 77 re-exported symbols). No state.py facade exists contrary to submission claim.
• All 4 specific import examples verified accurate.

Duplicate check: S034 (@xinlingfeiwu) — "app/ bypasses engine facades — 57 private imports" — verified YES_WITH_CAVEATS (5/6/4/5). Same core finding. S034 correctly identified that only plan.py facade exists.

Fix

No fix needed — verdict is NO (duplicate of S034).

Verdict

Question	Answer	Reasoning
Is this poor engineering?	YES	Private module boundary violations are real — 87 imports bypass underscore convention across 4 subpackages
Is this at least somewhat significant?	YES	Refactoring engine internals would break 55 consumer files across two architectural layers

Final verdict: NO

Duplicate of S034, which was submitted earlier and provided more accurate analysis (correctly identified only plan.py facade exists; S235 fabricates a state.py facade).

Scores

Criterion	Score
Significance	5/10
Originality	0/10
Core Impact	4/10
Overall	0/10

Summary

S235 identifies the same private module boundary violations as S034 — app/ and intelligence/ bypassing engine._* private subpackages. The submission broadens scope to include intelligence/ (30 additional imports for 87 total vs S034's 57) but the core architectural observation is identical. Two factual errors: claims a state.py facade exists (only plan.py does) and undercounts StateModel imports (24 vs actual 27). S034 submitted earlier with more accurate facade analysis.

Why Desloppify Missed This

• What should catch: An import-boundary or layering detector scanning for cross-layer private module access
• Why not caught: Desloppify analyzes code quality within files/functions, not cross-module import graph violations
• What could catch: import-linter (already configured in .github/importlinter.ini) or a custom detector checking for from pkg._private patterns across layer boundaries

Verdict Files

• Verdict JSON
• Verdict Report

Generated with Lota