proffesor-for-testing__agentic-qe/docs/qe-reports-3-6-3/dependency-analysis.md

Dependency Analysis Report - Agentic QE v3.6.3

Generated: 2026-02-11 Scope: /workspaces/agentic-qe-new/v3/ Source files analyzed: 937 TypeScript files Top-level modules: 37 Domains: 13 bounded contexts

---

Executive Summary

The v3 codebase has significant architectural debt in its dependency structure. While the Domain-Driven Design intent is clear (13 bounded contexts, microkernel architecture, shared kernel), the actual import graph reveals 88 circular dependency chains at the module level, 8 upward dependency violations in the shared kernel, and every domain violating its boundary by importing directly from coordination, integrations, and other infrastructure modules.

Critical findings:

1. Shared kernel (shared/) imports upward into learning/ (7 files) and mcp/ (1 file), breaking the foundational layer.
2. Kernel imports all 13 domain plugins directly (hard-coded factory map), creating a massive downward coupling that defeats the plugin architecture's purpose.
3. MCP layer has the highest instability (I=0.83) with fan-out to 25 modules -- it acts as a "god module" that reaches into everything.
4. All 13 domains bypass their bounded context by importing directly from coordination/, integrations/, and sibling domain barrel files.
5. 1 high-severity vulnerability found: @isaacs/brace-expansion (transitive, fixable via npm audit fix).
6. 335 dynamic imports undermine tree-shaking and increase bundle unpredictability.
7. @faker-js/faker is in production dependencies -- a test data generator that adds 5.2MB to the production bundle.

---

1. External Dependency Analysis

1.1 Runtime Dependencies Inventory

Package	Version	Purpose	Size	Risk	Notes
better-sqlite3	^12.5.0	Core persistence (unified DB)	12MB	HIGH	Native C++ addon, requires node-gyp build toolchain
hnswlib-node	^3.0.0	HNSW vector search for embeddings	1.1MB	HIGH	Native C++ addon, platform-specific binaries
@xenova/transformers	^2.17.2	ML inference for embeddings	154MB	HIGH	Enormous footprint; downloads models at runtime
@ruvector/gnn	0.1.19	Graph neural networks	15MB (combined)	HIGH	Native Rust addon, pinned version, platform overrides needed
@ruvector/attention	0.1.3	Attention mechanisms	(in @ruvector)	HIGH	Native Rust addon, pinned version
@ruvector/sona	0.1.5	Rust neural addon	(in @ruvector)	MEDIUM	Native Rust addon
prime-radiant-advanced-wasm	^0.1.3	WASM agent booster	small	MEDIUM	WebAssembly module, less portable
pg	^8.17.2	PostgreSQL client (cloud sync)	small	MEDIUM	Network dependency; only needed for cloud sync
vibium	^0.1.2	Browser automation	small	MEDIUM	Integration-only, not core
axe-core	^4.11.1	Accessibility testing engine	2.9MB	LOW	Domain-specific (visual-accessibility)
@faker-js/faker	^10.2.0	Test data generation	5.2MB	MEDIUM	Should be devDependency -- test utility in prod deps
uuid	^9.0.0	UUID generation	tiny	LOW	Stable, widely used. Note: v13.0.0 available (major)
chalk	^5.6.2	Terminal colors	tiny	LOW	ESM-only since v5
commander	^12.1.0	CLI framework	tiny	LOW	v14 available (major bump)
fast-glob	^3.3.3	File globbing	tiny	LOW	Stable
fast-json-patch	^3.1.1	JSON patching	tiny	LOW	Stable
jose	^6.1.3	JWT/JWE/JWS	tiny	LOW	Security library, keep updated
ora	^9.0.0	CLI spinners	tiny	LOW	v9.3.0 available
cli-progress	^3.12.0	Progress bars	tiny	LOW	Stable
secure-json-parse	^4.1.0	Safe JSON parsing	tiny	LOW	Security library
typescript	^5.9.3	TypeScript compiler	moderate	LOW	Runtime use for test-generation AST analysis
yaml	^2.8.2	YAML parsing	tiny	LOW	Stable

1.2 Dev Dependencies

Package	Version	Purpose
@types/better-sqlite3	^7.6.13	Type definitions
@types/node	^20.10.0	Node.js type definitions (v25 available -- major gap)
@types/pg	^8.16.0	PostgreSQL type definitions
@types/uuid	^10.0.0	UUID type definitions
@types/ws	^8.18.1	WebSocket type definitions
@vitest/coverage-v8	^4.0.16	V8 coverage provider
esbuild	^0.27.2	Bundle builder for CLI/MCP
glob	^13.0.0	File matching
msw	^2.12.7	Mock Service Worker for tests
tsx	^4.21.0	TypeScript execution
vitest	^4.0.16	Test framework

1.3 Optional Dependencies

Package	Version	Purpose
@claude-flow/browser	^3.0.0	Browser integration
@claude-flow/guidance	^3.0.0-alpha.1	Guidance integration
@ruvector/attention-darwin-arm64	0.1.3	macOS ARM native binary
@ruvector/attention-linux-arm64-gnu	0.1.3	Linux ARM native binary
@ruvector/gnn-darwin-arm64	0.1.19	macOS ARM native binary
@ruvector/gnn-linux-arm64-gnu	0.1.19	Linux ARM native binary

1.4 Security Vulnerabilities

Severity	Package	Advisory	Status
HIGH	@isaacs/brace-expansion@5.0.0	GHSA-7h2j-956f-4vf2 - Uncontrolled Resource Consumption	Fix available via npm audit fix

1.5 Outdated Packages

Package	Current	Latest	Gap
@faker-js/faker	10.2.0	10.3.0	Patch
@ruvector/attention	0.1.3	0.1.4	Patch
@ruvector/gnn	0.1.19	0.1.22	Patch
@types/node	20.19.30	25.2.3	Major
commander	12.1.0	14.0.3	Major
ora	9.0.0	9.3.0	Minor
pg	8.17.2	8.18.0	Patch
uuid	9.0.1	13.0.0	Major
vibium	0.1.4	0.1.7	Patch

1.6 Dependency Overrides (Supply Chain Risk)

The project uses overrides in package.json to remap musl-based Rust native binaries to gnu equivalents:

"overrides": {
    "@ruvector/gnn-linux-x64-musl": "npm:@ruvector/gnn-linux-x64-gnu@0.1.19",
    "@ruvector/gnn-linux-arm64-musl": "npm:@ruvector/gnn-linux-arm64-gnu@0.1.19",
    "tar": ">=7.5.7"
}

This is a fragile pattern -- if the musl/gnu ABIs diverge, runtime crashes will occur silently.

1.7 Native Module Installation Matrix

Package	Linux x64	Linux ARM64	macOS ARM64	Windows	Alpine (musl)
better-sqlite3	Build	Build	Build	Build	Build
hnswlib-node	Build	Build	Build	Build	Build
@ruvector/gnn	Prebuilt	Optional	Optional	N/A	Override to gnu
@ruvector/attention	Prebuilt	Optional	Optional	N/A	N/A

---

2. Internal Module Dependency Graph

2.1 Module-Level Coupling Metrics

Module	Ca (Afferent)	Ce (Efferent)	I (Instability)	Files	Risk
shared	30	2	0.06	61	MEDIUM (high Ca = fragile)
kernel	28	17	0.38	14	MEDIUM (high Ce for core)
integrations	22	4	0.15	113	MEDIUM (high Ca = fragile)
coordination	18	13	0.42	88	MEDIUM
learning	17	7	0.29	28	MEDIUM
domains (barrel)	16	16	0.50	3	MEDIUM
mcp	5	25	0.83	89	HIGH (god module)
domains/visual-accessibility	5	6	0.55	15	MEDIUM
domains/code-intelligence	5	5	0.50	15	LOW
domains/learning-optimization	5	6	0.55	9	MEDIUM
optimization	4	3	0.43	8	LOW
routing	4	1	0.20	9	LOW
memory	3	2	0.40	10	LOW
governance	3	1	0.25	16	LOW
adapters	3	4	0.57	74	MEDIUM
hooks	3	3	0.50	6	LOW
init	3	3	0.50	35	LOW
strange-loop	3	1	0.25	19	LOW
workers	1	6	0.86	17	MEDIUM
cli	0	13	1.00	50	MEDIUM (leaf)
index	0	15	1.00	1	MEDIUM (entry point)

Legend: Ca = afferent coupling (who depends on me), Ce = efferent coupling (who I depend on), I = instability (Ce / (Ca+Ce)), closer to 1.0 = more unstable.

2.2 Top 30 High Fan-In Files (Fragile Points)

These files are depended upon by the most other files. Changes here have the widest blast radius.

Dependents	File	Risk
137	kernel/interfaces	CRITICAL
128	shared/types	CRITICAL
85	shared/types/index	CRITICAL
42	learning/qe-patterns	HIGH
39	mcp/types	HIGH
31	integrations/rl-suite/interfaces	HIGH
29	mcp/tools/base	HIGH
27	shared/value-objects/index	HIGH
20	kernel/unified-memory	HIGH
20	shared/llm/interfaces	HIGH
18	coordination/mincut/queen-integration	MEDIUM
17	domains/test-generation/interfaces	MEDIUM
17	coordination/mincut/interfaces	MEDIUM
17	cli/handlers/interfaces	MEDIUM
16	init/types	MEDIUM
16	shared/events/domain-events	MEDIUM
16	workers/interfaces	MEDIUM
15	learning/token-tracker	MEDIUM
15	init/phases/phase-interface	MEDIUM
15	domains/domain-interface	MEDIUM
15	coordination/mixins/mincut-aware-domain	MEDIUM
15	coordination/mixins/consensus-enabled-domain	MEDIUM
15	coordination/consensus/domain-findings	MEDIUM
15	shared/utils/vector-math	MEDIUM
14	domains/visual-accessibility/interfaces	LOW
14	integrations/embeddings/base/types	LOW
14	coordination/consensus/interfaces	LOW
14	governance/feature-flags	LOW
13	integrations/coherence/types	LOW
13	coordination/mixins/governance-aware-domain	LOW

2.3 Top 30 High Fan-Out Files (Complexity Points)

These files import the most other files. They are complex and hard to test in isolation.

Dependencies	File	Risk
39	cli/index.ts	HIGH
25	coordination/queen-coordinator.ts	HIGH
24	kernel/kernel.ts	HIGH
23	domains/code-intelligence/coordinator.ts	HIGH
21	mcp/tools/registry.ts	HIGH
20	learning/index.ts	HIGH
19	coordination/index.ts	HIGH
19	mcp/tools/index.ts	HIGH
18	index.ts	MEDIUM
18	domains/test-execution/coordinator.ts	MEDIUM
18	domains/quality-assessment/coordinator.ts	MEDIUM
17	domains/contract-testing/coordinator.ts	MEDIUM
17	domains/test-generation/coordinator.ts	MEDIUM
17	domains/security-compliance/coordinator.ts	MEDIUM
17	integrations/ruvector/index.ts	MEDIUM
16	domains/coverage-analysis/coordinator.ts	MEDIUM
16	domains/learning-optimization/coordinator.ts	MEDIUM
16	domains/requirements-validation/coordinator.ts	MEDIUM
15	domains/chaos-resilience/coordinator.ts	MEDIUM
15	domains/visual-accessibility/coordinator.ts	MEDIUM
15	coordination/mincut/index.ts	MEDIUM
15	governance/index.ts	MEDIUM
14	init/phases/index.ts	LOW
14	domains/index.ts	LOW
14	mcp/handlers/core-handlers.ts	LOW
14	cli/handlers/init-handler.ts	LOW
12	domains/defect-intelligence/coordinator.ts	LOW
12	coordination/task-executor.ts	LOW
12	shared/index.ts	LOW
12	cli/commands/fleet.ts	LOW

---

3. Circular Dependencies

3.1 Module-Level Circular Dependencies (Critical)

The analysis detected 88 circular dependency chains at the module level. Below are the fundamental cycles that generate all the transitive cycles:

#	Cycle	Severity	Root Cause
1	kernel <-> memory	CRITICAL	kernel/unified-memory imports memory/crdt; memory imports kernel
2	kernel <-> domains/* (all 13)	CRITICAL	kernel.ts hard-imports all domain plugin factories
3	coordination <-> domains/code-intelligence	HIGH	queen-coordinator imports domain; domain imports coordination mixins
4	coordination <-> domains/defect-intelligence	HIGH	Same pattern as above
5	shared <-> mcp	HIGH	shared/io/file-reader imports mcp/security/cve-prevention
6	shared <-> learning	HIGH	7 LLM providers import learning/token-tracker
7	learning <-> feedback	MEDIUM	feedback loop integration
8	learning <-> domains/coverage-analysis	MEDIUM	HNSW index usage
9	learning <-> adapters	MEDIUM	claude-flow adapter
10	integrations <-> kernel	MEDIUM	kernel imports via migrations; integrations import kernel
11	domains/* <-> domains (barrel)	MEDIUM	domain-interface.ts imports coordination types; domains/index.ts re-exports all

Impact: These cycles mean that changing any file in kernel/, shared/, or coordination/ can trigger recompilation of the entire codebase. They also prevent proper code splitting and lazy loading.

3.2 File-Level Circular Dependencies (Direct A<->B)

#	File A	File B	Module
1	coordination/queen-coordinator.ts	coordination/mincut/queen-integration.ts	coordination
2	learning/v2-to-v3-migration.ts	learning/qe-unified-memory.ts	learning
3	agents/claim-verifier/index.ts	agents/claim-verifier/verifiers/output-verifier.ts	agents
4	agents/claim-verifier/index.ts	agents/claim-verifier/verifiers/file-verifier.ts	agents
5	adapters/a2a/notifications/webhook-service.ts	adapters/a2a/notifications/subscription-store.ts	adapters

---

4. Domain Boundary Analysis

4.1 Domain Boundary Violations

Every domain violates its bounded context boundary by importing directly from infrastructure modules. In a proper DDD architecture, domains should only depend on shared/ (shared kernel) and kernel/ (interfaces). Communication between domains and infrastructure should happen through the event bus or dependency injection.

Domain	Violations (imports from)
chaos-resilience	coordination, integrations, mcp, domains(barrel)
code-intelligence	coordination, integrations, domains(barrel)
contract-testing	coordination, integrations, domains(barrel)
coverage-analysis	coordination, integrations, domains(barrel)
defect-intelligence	coordination, integrations, causal-discovery, domains(barrel)
enterprise-integration	coordination, domains(barrel)
learning-optimization	coordination, integrations, learning, domains(barrel)
quality-assessment	coordination, integrations, agents, domains(barrel)
requirements-validation	coordination, integrations, domains(barrel)
security-compliance	coordination, integrations, domains(barrel)
test-execution	coordination, integrations, strange-loop, domains(barrel)
test-generation	coordination, integrations, domains(barrel)
visual-accessibility	coordination, integrations, adapters, domains(barrel)

Common violations across all domains:

• coordination/: Every domain imports coordination/mixins/ (governance-aware-domain, consensus-enabled-domain, mincut-aware-domain). These mixins couple domains to the coordination infrastructure.
• integrations/: Domains import integrations/coherence/, integrations/rl-suite/, and integrations/embeddings/ directly instead of through injected interfaces.
• domains/ barrel: Domains import from their parent barrel file (../domain-interface.js, ../../constants.js), which is structurally acceptable but creates coupling to the barrel file's transitive dependencies.

4.2 Shared Kernel Upward Dependencies (Layering Violations)

The shared kernel (shared/) is supposed to be the lowest layer with zero upward dependencies. It currently violates this:

File	Imports From	Problem
shared/llm/providers/bedrock.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/ollama.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/azure-openai.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/openai.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/gemini.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/claude.ts	learning/token-tracker	Shared depends on learning
shared/llm/providers/openrouter.ts	learning/token-tracker	Shared depends on learning
shared/io/file-reader.ts	mcp/security/cve-prevention	Shared depends on MCP

Root cause: TokenMetricsCollector is defined in learning/ but is used by all 7 LLM providers in shared/. It should either live in shared/ or be injected via an interface.

4.3 Cross-Domain Communication Patterns

Pattern	Observed Usage	Conformance
Event Bus	kernel/event-bus.ts provides InMemoryEventBus	Correct but underutilized
Direct Import	All 13 domains import coordination mixins directly	Violation
Plugin DI	Kernel injects EventBus + MemoryBackend into domains	Correct
Shared Kernel Types	All domains import from shared/types	Correct
Coordination Mixins	Domains extend governance/consensus/mincut mixins	Architectural smell

4.4 External Package Usage by Domain

Domain	External Packages Used
chaos-resilience	uuid, net, child_process
code-intelligence	path, fs, uuid, child_process
contract-testing	uuid
coverage-analysis	uuid, fs, path
defect-intelligence	uuid
enterprise-integration	uuid
learning-optimization	uuid
quality-assessment	uuid
requirements-validation	uuid, path, fs
security-compliance	uuid, path, fs, child_process, util
test-execution	uuid, path, fs, child_process
test-generation	uuid, @faker-js/faker, fs, path, typescript, chai
visual-accessibility	uuid, fs, module

Concern: test-generation imports @faker-js/faker and chai -- test libraries that should not be in production domain code unless they are generating test files as output (which is the domain's purpose, so this is acceptable).

---

5. Build and Bundle Analysis

5.1 Module System Configuration

Target: ES2022
Module: ESNext
Module Resolution: bundler
ESM: Yes (package.json "type": "module")
Build: tsc + esbuild (CLI/MCP bundles)

The project is ESM-native, which is good for tree-shaking. However, the actual tree-shaking effectiveness is undermined by:

5.2 Tree-Shaking Risks

Barrel files with excessive star re-exports (prevent tree-shaking of unused exports):

File	Star Re-exports	Risk
shared/index.ts	11	HIGH -- everything in shared gets pulled in
index.ts (root)	8	HIGH -- root entry pulls everything
domains/test-execution/services/index.ts	8	MEDIUM
coordination/protocols/index.ts	6	MEDIUM
governance/index.ts	5	MEDIUM
coordination/index.ts	4	LOW
integrations/browser/agent-browser/index.ts	4	LOW

5.3 Dynamic Imports

335 dynamic imports/requires were detected. While some are intentional for lazy loading (e.g., better-sqlite3 in init wizard), many undermine static analysis:

• init/init-wizard.ts: 13 dynamic imports (acceptable -- lazy loading heavy deps during initialization)
• mcp/ handlers: Multiple dynamic imports for domain handlers (acceptable -- on-demand loading)
• integrations/: Dynamic require() calls for native modules (necessary for optional deps)

5.4 Bundle Size Impact (node_modules)

Component	Size	Notes
Total node_modules	865MB
@xenova/transformers	154MB	17.8% of total -- ML model inference
@ruvector/*	15MB	Native Rust binaries
better-sqlite3	12MB	Native C++
@faker-js/faker	5.2MB	Should be devDep
axe-core	2.9MB	Accessibility engine
hnswlib-node	1.1MB	Native C++

5.5 Side Effects

No bare side-effect imports (import "x") were found, which is positive for tree-shaking. However, the heavy use of barrel files with export * largely negates this benefit.

---

6. Architectural Risk Assessment

6.1 Dependency Depth (Critical Paths)

User Request
  -> mcp/server.ts
    -> mcp/tools/registry.ts (21 deps)
      -> kernel/kernel.ts (24 deps)
        -> domains/*/plugin.ts (each 15-18 deps)
          -> coordination/mixins/* (15 deps each)
            -> shared/types (128 dependents)
              -> learning/token-tracker (CYCLE!)

Maximum dependency chain depth from MCP entry point to leaf: 6-7 levels.

6.2 Stability vs. Abstractness (Zone Analysis)

Using Robert C. Martin's dependency metrics:

Zone	Modules	Assessment
Zone of Pain (stable + concrete)	shared/, kernel/interfaces	Too many dependents AND too concrete. Changes here are expensive.
Zone of Uselessness (abstract + unstable)	None detected	No purely abstract unused modules
Main Sequence (balanced)	coordination/, governance/, routing/	Appropriate balance
Zone of Instability (unstable + concrete)	mcp/, cli/, workers/	High fan-out, but acceptable as edge modules

6.3 Kernel Architecture Issues

The kernel (kernel/kernel.ts) hard-imports all 13 domain plugin factories:

import { createTestGenerationPlugin } from '../domains/test-generation/plugin';
import { createTestExecutionPlugin } from '../domains/test-execution/plugin';
// ... 11 more

This creates a static dependency from the kernel to every domain, defeating the microkernel's purpose of dynamic loading. The PluginLoader class exists and supports factory registration, but the kernel pre-imports everything.

Recommendation: Move plugin factory registration to a configuration file or dynamic import map. The kernel should discover plugins at runtime, not compile-time.

---

7. Recommendations

7.1 Critical (Address Immediately)

#	Issue	Fix	Effort
1	Run npm audit fix	Fix @isaacs/brace-expansion vulnerability	5 min
2	Move @faker-js/faker to devDependencies	npm install --save-dev @faker-js/faker and update test-generation domain imports	30 min
3	Move TokenMetricsCollector to shared	Extract interface to shared/metrics/ or shared/types/, keep implementation in learning/ and inject	2-4 hrs
4	Move validatePath to shared	Extract mcp/security/cve-prevention path validation to shared/security/	1 hr

7.2 High Priority (Plan for Next Sprint)

#	Issue	Fix	Effort
5	Break kernel -> domains cycle	Replace static imports in kernel.ts with dynamic plugin discovery or a registry pattern	4-8 hrs
6	Extract coordination mixins to interfaces	Create shared/coordination-interfaces/ with mixin interfaces; domains depend on interfaces, not implementations	8-16 hrs
7	Reduce MCP fan-out	Split mcp/ into sub-modules with clear facades; use lazy loading for domain handlers	8-16 hrs
8	Fix file-level circulars	5 direct A<->B cycles need extraction of shared types or mediator patterns	4-8 hrs

7.3 Medium Priority (Technical Debt Reduction)

#	Issue	Fix	Effort
9	Replace barrel export * with named exports	Especially in shared/index.ts (11 star re-exports) -- use explicit named exports for tree-shaking	4 hrs
10	Make @xenova/transformers optional	Lazy-load only when embedding features are used; add graceful fallback	4-8 hrs
11	Enforce domain boundary rules in CI	Add ESLint rule or custom linter to prevent domains from importing coordination/integrations directly	4 hrs
12	Update major dependencies	uuid@9 -> 13, commander@12 -> 14, @types/node@20 -> 25	4-8 hrs
13	Review 335 dynamic imports	Audit each for necessity; convert unnecessary dynamic imports to static for better static analysis	8 hrs

7.4 Long-Term Architecture Improvements

#	Improvement	Description
14	Domain event bus adoption	Domains should communicate exclusively through domain events, not direct imports
15	Proper Anti-Corruption Layers	Each domain should have an ACL for external integrations instead of importing integrations/ directly
16	Dependency injection container	Introduce a DI container at the kernel level to decouple all cross-module dependencies
17	Package-level modules	Consider splitting into npm workspaces: @aqe/kernel, @aqe/shared, @aqe/domains-* for hard boundary enforcement
18	Native dependency strategy	Document and test all platform combinations for native modules; consider pure-JS fallbacks

---

Appendix A: Full Module Dependency Edges

adapters -> integrations, kernel, learning, shared
agents -> shared
benchmarks -> shared
causal-discovery -> integrations, learning
cli -> coordination, domains/requirements-validation, domains/visual-accessibility,
       init, integrations, kernel, learning, mcp, migration, optimization, shared, sync, validation
coordination -> domains, domains/code-intelligence, domains/defect-intelligence,
               domains/learning-optimization, domains/security-compliance, governance,
               hooks, integrations, kernel, learning, mcp, routing, shared
domains/chaos-resilience -> coordination, domains, integrations, kernel, mcp, shared
domains/code-intelligence -> coordination, domains, integrations, kernel, shared
domains/contract-testing -> coordination, domains, integrations, kernel, shared
domains/coverage-analysis -> coordination, domains, integrations, kernel, shared
domains/defect-intelligence -> causal-discovery, coordination, domains, integrations, kernel, shared
domains/enterprise-integration -> coordination, domains, kernel, shared
domains/learning-optimization -> coordination, domains, integrations, kernel, learning, shared
domains/quality-assessment -> agents, coordination, domains, integrations, kernel, shared
domains/requirements-validation -> coordination, domains, integrations, kernel, shared
domains/security-compliance -> coordination, domains, integrations, kernel, shared
domains/test-execution -> coordination, domains, integrations, kernel, shared, strange-loop
domains/test-generation -> coordination, domains, integrations, kernel, shared
domains/visual-accessibility -> adapters, coordination, domains, integrations, kernel, shared
feedback -> learning, routing
governance -> shared
hooks -> learning, memory, types
init -> kernel, learning, optimization
integrations -> kernel, learning, migrations, shared
kernel -> coordination, domains/* (all 13), memory, migrations, shared
learning -> adapters, domains/coverage-analysis, feedback, governance,
           integrations, kernel, shared
mcp -> adapters, coordination, domains (multiple), governance, hooks, init,
      integrations, kernel, learning, memory, optimization, planning, routing,
      shared, strange-loop, types
memory -> kernel, types
optimization -> kernel, learning, workers
planning -> kernel, learning
routing -> learning
shared -> learning (!), mcp (!)
strange-loop -> integrations
sync -> integrations, kernel
test-scheduling -> domains/code-intelligence, kernel
validation -> learning
workers -> domains/learning-optimization, domains/quality-assessment,
          domains/test-execution, kernel, learning, shared

Appendix B: External Package Usage Map

Module	Packages
adapters	events, uuid, fs, path, jose, fast-json-patch, crypto
cli	chalk, commander, fs, path, readline, os, ora, cli-progress, secure-json-parse
coordination	uuid, crypto, fs, path
domains (all)	uuid (primary), fs, path, child_process (some)
init	fs, path, module, url, chalk, child_process
integrations	better-sqlite3, uuid, path, fs, child_process, @xenova/transformers, hnswlib-node, @ruvector/*
kernel	better-sqlite3, fs, path, uuid
learning	uuid, better-sqlite3, crypto
mcp	uuid, http, crypto, path, fs, events, ws
planning	better-sqlite3, crypto
sync	path, fs, secure-json-parse, better-sqlite3, uuid

---

Report generated by QE Dependency Mapper v3 -- 2026-02-11