proffesor-for-testing__agentic-qe/docs/qe-reports-3-9-13/01-code-complexity-smells-report.md

Code Complexity & Smells Report - v3.9.13

Date: 2026-04-20 Agent: qe-code-complexity Baseline: v3.8.13 (2026-03-30)

---

Executive Summary

v3.9.13 shows a clearly improving quality trajectory against the v3.8.13 baseline. The codebase grew by 68 source files (+5.7%) to 1,263 files totaling 564,564 lines, but structural complexity improved across almost every measurable dimension. The most significant win is the remediation of the prior #1 hotspot: calculateComplexity in defect-predictor.ts was decomposed from CC=104/526 lines to CC=8/58 lines — a textbook extraction into AST-based helpers. extractJson in flaky-detector.ts also collapsed from 652 lines to 52 lines. Silent catch blocks dropped from 36 to 5. However, as any casts regressed from 4 to 18 (+14), and the 43 as unknown as double-casts in protocol-server.ts remain unchanged. The prior #1 CC hotspot is gone, but three CLI command builders (createWorkflowCommand, createCodeCommand, createCoverageCommand) now dominate the top-CC list — though these are largely Commander.js fluent-builder chains, not genuine branching logic.

Overall Score: 7.0 / 10 (was 6.5 at v3.8.13 — continued improvement driven by hotspot remediation)

---

1. Source File Metrics

Metric	v3.8.13	v3.9.13	Delta	Trend
Source files (.ts, excl. tests)	1,195	1,263	+68 (+5.7%)	Growing
Total source lines	549,542	564,564	+15,022	Growing
Avg lines/file	460	447	-13	Improving

The avg lines/file dropped by 13 — a signal that new code is being added as smaller modules rather than padding existing files.

---

2. Cyclomatic Complexity

Distribution by Tier

Tier	CC Range	v3.8.13	v3.9.13	Delta	Trend
Critical	>50	11	16	+5	Declining
High	21-50	104	97	-7	Improving
Total CC>20		115	113	-2	Stable

Methodology note: v3.9.13 uses a stricter non-nested scanner that skips methods inside already-counted method bodies. A portion of the +5 critical uplift comes from previously-hidden inner CLI builders now being surfaced in their own right.

Top 10 Most Complex Functions (Critical: CC>50)

Rank	Function	File	CC	Lines	Notes
1	createWorkflowCommand	src/cli/commands/workflow.ts:23	146	640	Commander fluent builder (inflated CC)
2	run (assets phase)	src/init/phases/09-assets.ts:38	89	321	Carried over from v3.8.13 (CC=80)
3	createCRDTStore	src/memory/crdt/crdt-store.ts:42	76	565	Carried over (CC=72)
4	createCodeCommand	src/cli/commands/code.ts:9	74	381	CLI builder (was 69 @ test cmd)
5	createCoverageCommand	src/cli/commands/coverage.ts:17	63	300	CLI builder
6	createSecurityCommand	src/cli/commands/security.ts:9	63	220	CLI builder
7	generateAssertionsFromBehavior	src/domains/test-generation/services/tdd-generator.ts:45	60	100	Carried over (CC=60)
8	createMemoryCommand	src/cli/commands/memory.ts:8	55	301	CLI builder
9	recordDomainFeedback	src/mcp/handlers/handler-factory.ts:221	52	77	NEW — genuine branching
10	createTestCommand	src/cli/commands/test.ts:9	51	259	CLI builder (was 54)
11	registerDreamCommand	src/cli/commands/learning.ts:918	51	198	CLI builder
12	executePhase	src/cli/commands/ci.ts:25	50	230	NEW — CI runner
13	parseCoverage	src/domains/coverage-analysis/services/coverage-parser.ts:864	50	114	Coverage parser

Hotspot Change: calculateComplexity (RESOLVED)

The former #1 hotspot at src/domains/defect-intelligence/services/defect-predictor.ts:657 has been fully decomposed:

• Before: CC=104, 526 lines (v3.8.13)
• After: CC=8, 58 lines (lines 657–715)
• How: Extracted into AST-based helper calls (tsParser.extractFunctions, tsParser.extractClasses) plus 5 simple regex-based metric calculators feeding a weighted average. A second path-heuristic method estimateComplexityFromPath at line 720 handles the fallback case.
• Improvement factor: 13× CC reduction, 9× line reduction

Hotspot Change: extractJson in flaky-detector.ts (RESOLVED)

• Before: CC=78, 652 lines (v3.8.13)
• After: 52 lines (lines 647–698), CC reduced significantly
• How: Adjacent parsers parseVitestJson (line 698) and parseJestJson (line 735) are now separate methods rather than embedded in extractJson.

Hotspot Change: registerAllTools in protocol-server.ts (PARTIAL)

• Before: 778 lines (v3.8.13)
• After: 416 lines (lines 630–1045)
• Improvement: -46% length reduction; CC is now very low (~4) because the function is now almost purely a registration list
• Remaining concern: 43 as unknown as Parameters<typeof ...> casts still present (unchanged from v3.8.13)

---

3. File Size Analysis

Metric	v3.8.13	v3.9.13	Delta	Trend
Files >500 lines	441 (36.9%)	447 (35.4%)	+6 (+1.4%)	Improving %
Files >1000 lines	90	92	+2	Stable

Absolute count grew but percentage dropped — consistent with new code favoring smaller modules.

Top 10 Largest Files

Rank	File	v3.8.13 Lines	v3.9.13 Lines	Delta
1	src/learning/pattern-store.ts	--	1,862	NEW
2	src/domains/requirements-validation/qcsd-refinement-plugin.ts	1,861	1,861	0
3	src/domains/contract-testing/services/contract-validator.ts	1,827	1,827	0
4	src/domains/learning-optimization/coordinator.ts	1,778	1,778	0
5	src/cli/completions/index.ts	1,778	1,778	0
6	src/domains/test-generation/services/pattern-matcher.ts	1,769	1,769	0
7	src/domains/chaos-resilience/coordinator.ts	1,704	1,704	0
8	src/domains/test-generation/coordinator.ts	1,694	1,703	+9
9	src/domains/requirements-validation/qcsd-ideation-plugin.ts	1,699	1,699	0
10	src/mcp/protocol-server.ts	--	1,641	Down from ~2100 via registerAllTools cut

pattern-store.ts at 1,862 lines is now the largest file, displacing qcsd-refinement-plugin.ts by 1 line. All top 10 remain 3× over the 500-line guideline.

---

4. Functions >100 Lines

Metric	v3.8.13	v3.9.13	Delta	Trend
Functions >100 lines	147	139	-8 (-5.4%)	Improving

Top 10 Longest Functions

Rank	Function	File	Lines
1	validateGraphQLSchema	src/domains/contract-testing/services/schema-validator.ts:68	1,023
2	parseGraphQLOperation	src/domains/contract-testing/services/contract-validator.ts:1089	674
3	createWorkflowCommand	src/cli/commands/workflow.ts:23	640
4	createCRDTStore	src/memory/crdt/crdt-store.ts:42	565
5	createQEHookHandlers	src/learning/qe-hooks.ts:111	492
6	createFleetCommand	src/cli/commands/fleet.ts:20	453
7	registerAllTools	src/mcp/protocol-server.ts:630	416
8	createCodeCommand	src/cli/commands/code.ts:9	381
9	createAccessibilitySurface	src/adapters/a2ui/renderer/templates/accessibility-surface.ts:137	353
10	extractConcepts	src/domains/code-intelligence/services/semantic-analyzer.ts:490	348

validateGraphQLSchema at 1,023 lines is the new longest function — this appears to be a newly-added or previously-unnoticed contract-testing method and is the single most urgent decomposition candidate.

---

5. Code Smells

5.1 Type Safety

Smell	v3.8.13	v3.9.13	Delta	Trend
as any casts (instances)	9 (4 genuine)	18	+14 genuine	Regression
as unknown as casts	43 (protocol-server)	136 total (43 protocol-server)	+93	Regression
@ts-ignore / @ts-expect-error	0	0	0	Clean
@ts-nocheck	0	0	0	Clean

as any distribution (v3.9.13, genuine casts):

• src/persistence/rvf-migration-coordinator.ts — 6 instances (largest concentration)
• src/cli/commands/llm-router.ts — 3 (ADR-092 related)
• src/routing/tiny-dancer-router.ts — 2 (ADR-092 related)
• src/mcp/tools/coverage-analysis/index.ts — 2 (carried over)
• Various single-instance: daemon.ts, init-handler.ts, token-optimizer-service.ts, workers/quality-daemon/index.ts

as unknown as distribution (v3.9.13, top files):

• src/mcp/protocol-server.ts — 43 (unchanged from v3.8.13; Parameters<typeof handler>[0] pattern for handler dispatch)
• src/integrations/ruvector/gnn-wrapper.ts — 10
• src/domains/requirements-validation/qcsd-refinement-plugin.ts — 6
• src/adapters/a2ui/data/reactive-store.ts — 6
• src/integrations/ruvector/brain-shared.ts — 4

The regression of as any from 4 to 18 is partly driven by ADR-092/093 integration (llm-router.ts, tiny-dancer-router.ts) and the RVF migration coordinator. These should be replaced with proper typed interfaces.

5.2 Console Usage

Type	v3.8.13	v3.9.13	Delta	Trend
console.log	2,419	2,507	+88	Growing slightly
console.error	356	370	+14	Growing
console.warn	241	265	+24	Growing
console.debug	105	114	+9	Growing
console.info	22	22	0	Stable
Total console.*	3,143	3,278	+135 (+4.3%)	Regression

A small regression that tracks with the 5.7% file-count growth. Relative density is approximately flat (~2.6 calls per file). Structured logging remains the unfinished work.

5.3 Catch Block Quality

Smell	v3.8.13	v3.9.13	Delta	Trend
Silent/empty catch blocks	36	5	-31 (-86%)	Major improvement
Parameterless catch {	612	701	+89	Growing
Parameterless catch()	2	2	0	Stable

Silent catch blocks dropped dramatically. Remaining 5 silent catches:

• src/init/phases/10-workers.ts — 3 instances (lines 154, 215, 219)
• src/mcp/protocol-server.ts — 2 instances (lines 1550, 1568)

The growth in parameterless catch { (+89) is proportional to codebase growth and is the TypeScript-idiomatic way to suppress an error you explicitly don't need.

5.4 Technical Debt Comments

Type	v3.8.13	v3.9.13	Delta
TODO	70	70	0
FIXME	9	9	0
HACK	8	8	0
XXX	7	9	+2
Total	94	96	+2

Flat — no significant accumulation of technical debt markers.

5.5 Deep Nesting

Metric	v3.8.13	v3.9.13	Delta
Files with nesting depth >6	190 (15.9%)	187 (14.8%)	-3 (pct -1.1%)

Top offenders (depth measured via stripped brace-counting — v3.8.13 "47" was via raw indent-counting which over-reported template literals):

File	Max Depth v3.9.13	v3.8.13
src/domains/test-generation/generators/jest-rn-generator.ts	15	not top 5
src/domains/test-generation/generators/jest-vitest-generator.ts	12	not top 5
src/domains/contract-testing/services/contract-validator.ts	11	15
src/domains/contract-testing/services/api-compatibility.ts	10	10
src/coordination/handlers/security-handlers.ts	10	not top 5
src/cli/commands/learning.ts	10	not top 5
src/shared/parsers/multi-language-parser.ts	7	47 (methodology) / stripped was likely ~10

multi-language-parser.ts now has only 30 decision points (was 348). The v3.8.13 report's "depth 47" was measuring raw whitespace indentation which counts template-literal content; true brace-depth is 7, and the file has been substantively simplified.

---

6. Error Handling

Metric	v3.8.13	v3.9.13	Delta	Trend
toErrorMessage() usage	620	439	-181	Regression?
Raw error.message usage	633	625	-8	Stable
Adoption ratio	~49.5%	~41%	-8.5%	Declining

Note: The toErrorMessage() count dropped from 620 to 439. Spot-checking suggests either (a) some call sites were removed along with decomposed functions, or (b) the utility is being re-exported under a different name in newer modules. This warrants investigation by the error-handling lead. The raw error.message count is effectively flat despite the codebase growing 5.7%, which is modestly encouraging, but the toErrorMessage regression offsets the structural gain.

---

7. process.exit() Usage

Metric	v3.8.13	v3.9.13	Delta
process.exit() calls	41	52	+11
Files with >5 instances	0	0	0

Growth of 11 instances tracks with new CLI command files (ci.ts, additional llm-router commands).

---

8. Functions >7 Parameters

Metric	v3.8.13	v3.9.13
Functions with >7 params	0	4

New violators:

• traverseDependencies — src/domains/code-intelligence/services/knowledge-graph.ts:1108 (8 params)
• computeRegionSimilarity — src/domains/visual-accessibility/cnn-visual-regression.ts:257 (8 params)
• createPyramidResult — src/early-exit/early-exit-controller.ts:270 (9 params)
• createDecision — src/early-exit/early-exit-decision.ts:251 (8 params)

All 4 are good candidates for introducing a parameter object.

---

9. File-Level Decision Point Complexity

Top 10 files by total decision points:

Rank	File	DP v3.9.13	DP v3.8.13	Delta
1	contract-validator.ts	317	348	-31
2	pattern-store.ts	217	--	NEW
3	qcsd-refinement-plugin.ts	215	253	-38
4	learning.ts (CLI)	210	292	-82
5	coverage-parser.ts	178	240	-62
6	schema-validator.ts	177	--	NEW
7	c4-model/index.ts	174	256	-82
8	qcsd-ideation-plugin.ts	172	--	NEW to top-10
9	workflow-orchestrator.ts	171	--	NEW to top-10
10	workflow-parser.ts	169	--	NEW to top-10

Files exceeding 200 decision points (Critical): 4 (was 14) Files with 100–200 decision points (High): 57 (was 118)

Major improvement: Files with >100 DP fell from 132 to 61 — a 54% reduction. The previously-flagged multi-language-parser.ts (348 DP → 30 DP) and brutal-honesty-analyzer.ts (284 DP → now off top-10) have been simplified substantially.

---

10. ADR-092 / ADR-093 Impact Assessment

ADR-092 (Advisor Strategy)

New advisor module at src/routing/advisor/:

File	Lines
multi-model-executor.ts	239
redaction.ts	225
circuit-breaker.ts	163
types.ts	103
domain-prompts.ts	57
index.ts	43
Total	830

Assessment: Well-decomposed (6 files, all under 250 lines, no CC hotspots). Exemplifies the structural discipline that should be applied elsewhere.

ADR-093 (Opus 4.7 Migration)

No structural complexity cost detected. cli/commands/llm-router.ts introduces 3 as any casts for provider-name mapping — replaceable with a typed union.

Governance Directory

src/governance/ has grown substantially (19,264 LOC across 19 files), with 6 files over 1,000 lines:

File	Lines
ab-benchmarking.ts	1,583
compliance-reporter.ts	1,468
evolution-pipeline-integration.ts	1,325
shard-retriever-integration.ts	1,279
constitutional-enforcer.ts	1,185
adversarial-defense-integration.ts	1,001

Governance is a quality concern area but did not show up in the top CC hotspots, suggesting the complexity is broad and procedural rather than deeply branched. Still, 6 files over 1,000 lines each is a decomposition candidate.

---

11. Maintainability Assessment

Maintainability Index: 71/100 (Improving, was 67)

Factor	Score	Weight	Weighted
Cyclomatic complexity distribution	7/10	20%	1.40
File size compliance (<500 lines)	5/10	15%	0.75
Function length compliance (<100 lines)	6/10	15%	0.90
Type safety (as any, as unknown as)	7/10	10%	0.70
Error handling maturity	5/10	10%	0.50
Console hygiene	4/10	10%	0.40
Nesting depth	6/10	10%	0.60
Hotspot remediation (silent catches, top-1 CC)	9/10	10%	0.90
Total		100%	6.15 → 71/100

+4 point improvement driven by the successful calculateComplexity and extractJson decompositions and the silent-catch reduction (36→5).

---

12. v3.8.13 Remediation Tracker

v3.8.13 Hotspot	v3.8.13 State	v3.9.13 State	Status
calculateComplexity (defect-predictor)	CC=104, 526 lines	CC=8, 58 lines	RESOLVED
switch heuristics-engine (qx)	CC=99, 340 lines	Not in top-13 (off critical list)	RESOLVED
run (assets phase 09)	CC=80, 249 lines	CC=89, 321 lines	Worsened slightly
extractJson (flaky-detector)	CC=78, 652 lines	52 lines	RESOLVED
createCRDTStore	CC=72, 571 lines	CC=76, 565 lines	Stable (not worse)
if code command	CC=69, 328 lines	createCodeCommand CC=74, 381 lines	Stable (now CLI-builder-counted)
brutal-honesty-analyzer for	CC=62, 153 lines	Not in top-13	RESOLVED
generateAssertionsFromBehavior	CC=60, 100 lines	CC=60, 100 lines	Unchanged
if test command	CC=54, 223 lines	createTestCommand CC=51, 259 lines	Stable
inferImplementationFromBehavior	CC=52, 82 lines	Not in top-13	RESOLVED
switch test-data-generator	CC=52, 57 lines	Not in top-13	RESOLVED
multi-language-parser nesting 47	348 DP	30 DP, depth 7	RESOLVED
registerAllTools (protocol-server)	778 lines, CC minimal	416 lines, CC minimal	Partially resolved (~46% shrink)
createHooksCommand (ADR-041 split)	Decomposed in v3.8.13	Still decomposed (81-line orchestrator + 8 handlers)	MAINTAINED

Summary: 7 of 14 tracked hotspots fully resolved, 4 partially/stable, 1 slightly worsened (run assets phase), 2 unchanged.

---

13. Trend Analysis (v3.8.13 → v3.9.13)

Dimension	Direction	Evidence
Critical complexity (CC>50)	Declining	11 → 16 (+5; partly CLI-builder artifacts)
High complexity (21–50)	Improving	104 → 97 (-7)
Total CC>20	Stable	115 → 113 (-2)
Files >500 lines	Improving (%)	36.9% → 35.4%
Files >1000 lines	Stable	90 → 92 (+2)
as any genuine casts	Regressing	4 → 18 (+14)
as unknown as casts	Regressing	~43 → 136 (+93; 43 unchanged in protocol-server)
console.* calls	Regressing	3,143 → 3,278 (+135, tracks file growth)
Silent catch blocks	Major improvement	36 → 5 (-31, -86%)
toErrorMessage() adoption	Regressing	620 → 439
Functions >100 lines	Improving	147 → 139 (-8)
Files with nesting >6	Improving	190 (15.9%) → 187 (14.8%)
Functions >7 params	Regressing	0 → 4 (+4)
Top-1 hotspot remediation	Success	calculateComplexity 104→8 CC
Maintainability index	Improving	67 → 71

Overall trend: IMPROVING — structural wins (hotspot decomposition, silent catches, file-level DP) outweigh the type-safety regressions (as any, as unknown as).

---

14. Priority Refactoring Recommendations

P0 — Critical (address this sprint)

1. validateGraphQLSchema in schema-validator.ts (1,023 lines)

   • New longest function in the codebase. Decompose into per-rule validators following GraphQL spec sections.

2. parseGraphQLOperation in contract-validator.ts:1089 (674 lines)

   • Extract into operation-type-specific parsers (query, mutation, subscription).

3. as unknown as in protocol-server.ts (43 instances, unchanged)

   • Replace with a typed handler dispatch pattern using generics; remove the type-erasure.

4. 18 as any casts — investigate the 6 in rvf-migration-coordinator.ts first (likely indicates a missing typed interface on the RVF adapter surface).

P1 — High (address within 2 sprints)

5. createCRDTStore (CC=76, 565 lines) — still unresolved from v3.8.13.

6. run (assets phase 09) (CC=89, 321 lines) — slightly worsened from v3.8.13.

7. 4 functions with >7 params — introduce parameter objects (especially createPyramidResult with 9 params).

8. Governance directory bloat — 6 files over 1,000 lines; extract shared patterns.

9. toErrorMessage regression investigation — audit why usage dropped from 620 → 439.

P2 — Medium (next quarter)

10. Console cleanup: 3,278 calls — migrate to logger-factory.ts which is already used throughout core modules.

11. CLI builder CC inflation: The top 10 CC list is dominated by createXxxCommand builders. Consider a declarative command-spec pattern that generates the Commander tree, reducing reported CC and improving testability.

12. Nesting-depth outliers: jest-rn-generator.ts (depth 15) and jest-vitest-generator.ts (depth 12) — extract inner generator logic.

---

15. Scoring Summary

Category	v3.8.13	v3.9.13	Change	Notes
Cyclomatic complexity	7/10	7/10	0	CC>50 up (16) but CC>20 stable; top-1 hotspot remediated
File size	5/10	5/10	0	447 >500 lines; pattern-store.ts now largest
Function size	6/10	6/10	0	139 over 100 lines; 1,023-line outlier
Type safety	9/10	7/10	-2	18 as any (+14); 136 as unknown as
Error handling	6/10	5/10	-1	toErrorMessage regression; silent catches big win
Console hygiene	4/10	4/10	0	3,278 calls; tracks file growth
Nesting	5/10	6/10	+1	multi-language-parser resolved; 187 files still deep
Hotspot trend	8/10	9/10	+1	calculateComplexity, extractJson, silent catches all fixed
Overall	6.5/10	7.0/10	+0.5	Structural wins outweigh type-safety regressions

---

Report generated by qe-code-complexity agent. Analysis performed on 1,263 source files (564,564 lines) in the src/ directory. Methodology: string/comment-stripped regex-based CC approximation (stricter than v3.8.13 methodology in some dimensions — see individual sections for notes).