Skip to content

_GRAPH-INDEX.md

Latest commit

 

History

History
375 lines (303 loc) · 16.4 KB

_GRAPH-INDEX.md

File metadata and controls

375 lines (303 loc) · 16.4 KB

bkit Graph Index

Obsidian graph view central hub. All components connect from this file.

v1.2.0 Refactoring: Skills consolidated, .claude/ removed from repo, single source of truth at root level

v1.2.1 Multi-Language Support: Extension-based file detection, 20+ language support, configurable patterns

v1.2.1 Language Tier System: 4-tier language classification reflecting AI-Native development and Vibe Coding trends

v1.2.3 SessionStart Enhancement: AskUserQuestion guidance with 4 options for session initialization

v1.3.0 Check-Act Iteration Loop: Automatic gap analysis and fix cycles with pdca-iterator agent

v1.3.1 Cross-Platform: All hooks converted from Bash (.sh) to Node.js (.js) for Windows/Mac/Linux compatibility

v1.4.0 Dual Platform Support: Claude Code + Gemini CLI simultaneous support, 80+ lib/common.js functions, 8-language Intent Detection

v1.4.1 Context Engineering: Optimal token curation perspective for LLM reasoning, Response Report Rule

Philosophy (4)

Context Engineering (NEW)

bkit is a practical implementation of Context Engineering:

┌─────────────────────────────────────────────────────────────────┐
│              bkit Context Engineering Components                 │
├─────────────────────────────────────────────────────────────────┤
│  Domain Knowledge (21 Skills)  → Structured domain knowledge     │
│  Behavioral Rules (11 Agents)  → Role-based behavioral rules     │
│  State Management (lib/common) → State management 76+ functions  │
│  5-Layer Hook System           → Context injection timing ctrl   │
│  Dynamic Injection             → Conditional context selection   │
└─────────────────────────────────────────────────────────────────┘

Philosophy

Core design principles and methodology:

  • [[philosophy/core-mission|core-mission]] - Core mission & 3 philosophies (Automation First, No Guessing, Docs=Code)
  • [[philosophy/ai-native-principles|ai-native-principles]] - AI-Native development & 3 core competencies
  • [[philosophy/pdca-methodology|pdca-methodology]] - PDCA cycle & 9-stage pipeline relationship

Skills (21)

Core Skills (2)

  • [[../skills/bkit-rules/SKILL|bkit-rules]] - PDCA rules + auto-triggering + code quality standards
  • [[../skills/bkit-templates/SKILL|bkit-templates]] - Document templates for consistent PDCA documentation

Level Skills (3)

  • [[../skills/starter/SKILL|starter]] - Starter level (static web, HTML/CSS/JS, Next.js basics)
  • [[../skills/dynamic/SKILL|dynamic]] - Dynamic level (BaaS fullstack with bkend.ai)
  • [[../skills/enterprise/SKILL|enterprise]] - Enterprise level (MSA/K8s/Terraform, AI Native)

Pipeline Phase Skills (10)

  • [[../skills/development-pipeline/SKILL|development-pipeline]] - 9-stage pipeline overview
  • [[../skills/phase-1-schema/SKILL|phase-1-schema]] - Schema/terminology definition
  • [[../skills/phase-2-convention/SKILL|phase-2-convention]] - Coding conventions
  • [[../skills/phase-3-mockup/SKILL|phase-3-mockup]] - Mockup development
  • [[../skills/phase-4-api/SKILL|phase-4-api]] - API design/implementation
  • [[../skills/phase-5-design-system/SKILL|phase-5-design-system]] - Design system
  • [[../skills/phase-6-ui-integration/SKILL|phase-6-ui-integration]] - UI implementation + API integration
  • [[../skills/phase-7-seo-security/SKILL|phase-7-seo-security]] - SEO/Security
  • [[../skills/phase-8-review/SKILL|phase-8-review]] - Code review + quality analysis
  • [[../skills/phase-9-deployment/SKILL|phase-9-deployment]] - Deployment

Specialized Skills (3)

  • [[../skills/zero-script-qa/SKILL|zero-script-qa]] - Zero Script QA (log-based testing)
  • [[../skills/mobile-app/SKILL|mobile-app]] - Mobile app development (React Native, Flutter)
  • [[../skills/desktop-app/SKILL|desktop-app]] - Desktop app development (Electron, Tauri)

Removed Skills (v1.2.0)

The following skills were consolidated:

  • task-classification → lib/common.js
  • level-detection → lib/common.js
  • pdca-methodology → bkit-rules
  • document-standards → bkit-templates
  • evaluator-optimizer → /pdca-iterate command
  • analysis-patterns → bkit-templates
  • ai-native-development → enterprise
  • monorepo-architecture → enterprise

Agents (11)

Level-Based Agents

  • [[../agents/starter-guide|starter-guide]] - Starter level guide (beginners)
  • [[../agents/bkend-expert|bkend-expert]] - Dynamic level (BaaS expert)
  • [[../agents/enterprise-expert|enterprise-expert]] - Enterprise level (CTO-level advisor)
  • [[../agents/infra-architect|infra-architect]] - Infrastructure architect (AWS/K8s/Terraform)

Task-Based Agents

  • [[../agents/pipeline-guide|pipeline-guide]] - Pipeline guide (9-phase development)
  • [[../agents/gap-detector|gap-detector]] - Gap analysis (design vs implementation)
  • [[../agents/design-validator|design-validator]] - Design validation
  • [[../agents/code-analyzer|code-analyzer]] - Code quality analysis
  • [[../agents/qa-monitor|qa-monitor]] - QA monitoring (Zero Script QA)
  • [[../agents/pdca-iterator|pdca-iterator]] - Iteration optimizer (Evaluator-Optimizer pattern)
  • [[../agents/report-generator|report-generator]] - Report generation

Skills - User Invocable (v1.4.5)

Note: Commands deprecated in v1.4.4+. Use Skills instead.

PDCA Skill (Unified)

  • /pdca plan - Create plan document
  • /pdca design - Create design document
  • /pdca do - Implementation guide
  • /pdca analyze - Run gap analysis
  • /pdca iterate - Auto-fix with Evaluator-Optimizer
  • /pdca report - Generate completion report
  • /pdca status - Show PDCA dashboard
  • /pdca next - Guide next PDCA step

Level Skills

  • /starter - Starter level project guidance
  • /dynamic - Dynamic level project guidance
  • /enterprise - Enterprise level project guidance

Pipeline Skills

  • /development-pipeline start - Start pipeline guide
  • /development-pipeline next - Next pipeline phase
  • /development-pipeline status - Pipeline progress

Utility Skills

  • /zero-script-qa - Run Zero Script QA
  • /claude-code-learning - Learning curriculum
  • /code-review - Code review and quality analysis

Hooks (3 events)

Global Hooks (hooks/hooks.json)

  • [[components/hooks/_hooks-overview|SessionStart]] - Plugin initialization with AskUserQuestion guidance

Skill Frontmatter Hooks

  • [[components/hooks/_hooks-overview|PreToolUse]] - Before Write/Edit operations (defined in SKILL.md)
  • [[components/hooks/_hooks-overview|PostToolUse]] - After Write operations (defined in SKILL.md)

Scripts (26)

Note: All scripts converted to Node.js (.js) in v1.3.1 for cross-platform compatibility

v1.4.0: Added 5 new phase completion handlers

Core Scripts (3)

  • scripts/pre-write.js - Unified PreToolUse hook (PDCA + classification + convention)
  • scripts/pdca-post-write.js - PostToolUse guidance after Write
  • scripts/select-template.js - Template selection by level

Phase Scripts (11)

  • scripts/phase-transition.js - PDCA phase transition validation (v1.4.0)
  • scripts/phase1-schema-stop.js - Schema phase completion (v1.4.0)
  • scripts/phase2-convention-pre.js - Convention check before write
  • scripts/phase2-convention-stop.js - Convention phase completion (v1.4.0)
  • scripts/phase3-mockup-stop.js - Mockup phase completion (v1.4.0)
  • scripts/phase4-api-stop.js - Zero Script QA after API implementation
  • scripts/phase5-design-post.js - Design token verification
  • scripts/phase6-ui-post.js - Layer separation verification
  • scripts/phase7-seo-stop.js - SEO/Security phase completion (v1.4.0)
  • scripts/phase8-review-stop.js - Review completion guidance
  • scripts/phase9-deploy-pre.js - Deployment environment validation

QA Scripts (3)

  • scripts/qa-pre-bash.js - QA setup before Bash
  • scripts/qa-monitor-post.js - QA completion guidance
  • scripts/qa-stop.js - QA session cleanup

Agent Scripts (5)

  • scripts/design-validator-pre.js - Design document validation
  • scripts/gap-detector-post.js - Gap analysis guidance
  • scripts/gap-detector-stop.js - Gap detector completion
  • scripts/iterator-stop.js - Iterator completion
  • scripts/analysis-stop.js - Analysis completion guidance

Utility Scripts (4)

  • scripts/pdca-pre-write.js - PDCA pre-write checks
  • scripts/archive-feature.js - Feature archiving
  • scripts/sync-folders.js - Folder synchronization
  • scripts/validate-plugin.js - Plugin validation

Infrastructure

Shared Library

  • lib/common.js - Shared utility functions (v1.4.0 Node.js, 80+ functions)

Platform Detection (v1.4.0)

  • detectPlatform() - Detect current platform ('claude' | 'gemini' | 'unknown')
  • isClaudeCode() - Check if running in Claude Code
  • isGeminiCli() - Check if running in Gemini CLI
  • getPluginPath() - Get plugin root path for current platform
  • getBkitConfig() - Load bkit.config.json with caching

Caching System (v1.4.0)

  • _cache - In-memory TTL-based cache object
  • TTL-based invalidation for config, status, and feature data

Debug Logging (v1.4.0)

  • debugLog() - Debug logging with platform-specific paths
  • Writes to ~/.claude/bkit-debug.log or ~/.gemini/bkit-debug.log

PDCA Status v2.0 (v1.4.0)

  • createInitialStatusV2() - Create PDCA Status v2.0 schema
  • migrateStatusToV2() - Auto-migrate from v1.0 schema
  • getDefaultFeatureStatus() - Get default status object for a feature

Multi-Feature Management (v1.4.0)

  • setActiveFeature() - Set current working feature
  • addActiveFeature() - Add new feature to tracking
  • getActiveFeatures() - Get all tracked features
  • switchFeatureContext() - Switch between feature contexts
  • getFeatureContext() - Get context for specific feature

Intent Detection (v1.4.0)

  • detectNewFeatureIntent() - Detect new feature request from user message
  • matchImplicitAgentTrigger() - Match message to agent trigger keywords
  • matchImplicitSkillTrigger() - Match message to skill trigger keywords
  • 8-language support: EN, KO, JA, ZH, ES, FR, DE, IT

Ambiguity Detection (v1.4.0)

  • calculateAmbiguityScore() - Calculate ambiguity in user request
  • generateClarifyingQuestions() - Generate AskUserQuestion options
  • detectAmbiguousTerms() - Find unclear terms in message

Requirement Tracking (v1.4.0)

  • extractRequirementsFromPlan() - Parse requirements from plan document
  • calculateRequirementFulfillment() - Calculate completion percentage
  • getUnfulfilledRequirements() - List incomplete requirements

Phase Validation (v1.4.0)

  • checkPhaseDeliverables() - Check required deliverables for phase
  • validatePdcaTransition() - Validate phase transition is allowed
  • getPhaseRequirements() - Get requirements for specific phase

Configuration (existing)

  • getConfig() - Read from bkit.config.json
  • getConfigArray() - Get array value from config

File Classification (existing)

  • isSourceFile() - Negative pattern + extension detection (30+ extensions)
  • isCodeFile() - Tier-based code file detection
  • isUiFile() - UI component files (.tsx, .jsx, .vue, .svelte, .astro)
  • isEnvFile() - Environment file detection

Language Tier System (existing)

  • getLanguageTier() - Get tier (1-4, experimental, unknown) for file
  • getTierDescription() - Get tier description (AI-Native, Mainstream, etc.)
  • getTierPdcaGuidance() - Get PDCA guidance based on tier
  • isTier1(), isTier2(), isTier3(), isTier4() - Tier check helpers

Feature Detection (existing)

  • extractFeature() - Multi-language feature extraction
  • findDesignDoc() - Find design document for feature
  • findPlanDoc() - Find plan document for feature

Task Classification (existing)

  • classifyTask(), classifyTaskByLines() - Task size classification
  • detectLevel() - Project level detection (Starter/Dynamic/Enterprise)
  • getPdcaGuidance() - Get PDCA guidance for task size

JSON Output Helpers (existing)

  • outputAllow(), outputBlock(), outputEmpty() - Hook response helpers
  • readStdinSync(), parseHookInput() - Hook input helpers

PDCA Task System (existing)

  • PDCA_PHASES - PDCA phase definitions constant
  • getPdcaTaskMetadata() - Generate task metadata
  • generatePdcaTaskSubject() - Generate task subject
  • generatePdcaTaskDescription() - Generate task description
  • generateTaskGuidance() - Generate task creation guidance
  • getPreviousPdcaPhase() - Get previous PDCA phase
  • findPdcaStatus() - Read docs/.pdca-status.json
  • getCurrentPdcaPhase() - Get current PDCA phase for feature

Language Tier System (v1.2.1)

bkit supports languages and frameworks organized by tier:

Tier Category Languages/Frameworks
Tier 1 AI-Native Essential Python, TypeScript, JavaScript, React/Next.js, Svelte
Tier 2 Mainstream Recommended Go, Rust, Dart, Vue, Astro, Flutter, Tauri
Tier 3 Domain Specific Java, Kotlin, Swift, C/C++, Angular, Electron
Tier 4 Legacy/Niche PHP, Ruby, C#, Scala, Elixir
Experimental Future Consideration Mojo, Zig, V

Tier Selection Criteria:

  • AI tool ecosystem compatibility (Copilot, Claude, Cursor)
  • Vibe Coding optimization
  • Market share (IEEE Spectrum 2025)
  • Training data availability

Configurable Patterns (v1.2.1)

  • BKIT_EXCLUDE_PATTERNS - Exclude directories (node_modules, pycache, .git, etc.)
  • BKIT_FEATURE_PATTERNS - Feature directory patterns (features, modules, packages, etc.)

Configuration

  • bkit.config.json - Centralized configuration
    • Task classification thresholds
    • Level detection rules
    • PDCA document paths
    • Naming conventions

Gemini CLI Support (v1.4.0)

bkit supports Gemini CLI as a secondary platform:

Item Claude Code Gemini CLI
Manifest .claude-plugin/plugin.json gemini-extension.json
Context File CLAUDE.md GEMINI.md
Commands commands/*.md (Markdown) commands/gemini/*.toml (TOML)
Hook Events PreToolUse / PostToolUse BeforeTool / AfterTool
Plugin Root $CLAUDE_PLUGIN_ROOT $extensionPath
Project Dir $CLAUDE_PROJECT_DIR $GEMINI_PROJECT_DIR

Hook Mapping:

Claude Code         →  Gemini CLI
─────────────────────────────────
SessionStart        →  SessionStart
PreToolUse          →  BeforeTool
PostToolUse         →  AfterTool
Stop                →  AgentStop

Shared Components (Cross-platform):

  • skills/ - All 21 skills work on both platforms
  • agents/ - All 11 agents work on both platforms
  • scripts/ - All 26 scripts use Node.js (cross-platform)
  • lib/common.js - Platform detection via detectPlatform()
  • templates/ - All 20 templates work on both platforms

Templates (20)

PDCA Templates

  • plan.template.md - Plan phase
  • design.template.md - Design phase
  • design-starter.template.md - Starter-level design
  • design-enterprise.template.md - Enterprise-level design
  • analysis.template.md - Gap analysis
  • report.template.md - Completion report
  • iteration-report.template.md - Iteration report

Pipeline Templates (10)

  • pipeline/phase-1-schema.template.md
  • pipeline/phase-2-convention.template.md
  • pipeline/phase-3-mockup.template.md
  • pipeline/phase-4-api.template.md
  • pipeline/phase-5-design-system.template.md
  • pipeline/phase-6-ui.template.md
  • pipeline/phase-7-seo-security.template.md
  • pipeline/phase-8-review.template.md
  • pipeline/phase-9-deployment.template.md
  • pipeline/zero-script-qa.template.md

Other Templates

  • CLAUDE.template.md - Project conventions
  • _INDEX.template.md - Document index

Triggers

  • [[triggers/trigger-matrix]] - Event-based trigger matrix
  • [[triggers/priority-rules]] - Priority and conflict rules

Scenarios

  • [[scenarios/scenario-write-code]] - Code write flow
  • [[scenarios/scenario-new-feature]] - New feature request
  • [[scenarios/scenario-qa]] - QA execution

Testing

  • [[testing/test-checklist]] - Test checklist