PDCA methodology + CTO-Led Agent Teams + AI coding assistant mastery for AI-native development
bkit-opencode is an OpenCode plugin that transforms how you build software with AI. It provides structured development workflows, automatic documentation, and intelligent code assistance through the PDCA (Plan-Do-Check-Act) methodology.
Also available for Claude Code: bkit-claude-code
Context Engineering is the systematic curation of context tokens for optimal LLM inference - going beyond simple prompt crafting to build entire systems that consistently guide AI behavior.
Traditional Prompt Engineering:
"The art of writing good prompts"
Context Engineering:
"The art of designing systems that integrate prompts, tools, and state
to provide LLMs with optimal context for inference"
bkit is a practical implementation of Context Engineering, providing a systematic context management system for OpenCode.
bkit implements Context Engineering through three interconnected layers:
| Layer | Components | Purpose |
|---|---|---|
| Domain Knowledge | 28 Skills | Structured expert knowledge (phases, levels, specialized domains) |
| Behavioral Rules | 17 Agents | Role-based constraints with model selection (opus/sonnet/haiku) |
| State Management | TypeScript Library | PDCA status, intent detection, ambiguity scoring, team coordination |
Context injection occurs through OpenCode's plugin hook system:
Hook 1: config → Agent/skill/MCP registration, permissions
Hook 2: event → PDCA init, level detection, team state
Hook 3: chat.message → Intent detection, skill/agent triggers
Hook 4: tool.execute.before → Skill activation, tool constraints
Hook 5: tool.execute.after → PDCA phase auto-advance, document tracking
Hook 6: experimental.chat.system.transform → PDCA status injection, next-step guidance
Hook 7: experimental.session.compacting → State preservation across context compaction
Hook 8: permission.ask → Dangerous command filtering
Learn more: See AI-NATIVE-DEVELOPMENT.md for detailed methodology.
- PDCA Methodology - Structured development workflow with automatic documentation
- CTO-Led Agent Teams - CTO agent orchestrates parallel PDCA execution with multi-agent teams (Dynamic: 3, Enterprise: 5 teammates)
- Plan Plus Skill - Brainstorming-enhanced PDCA planning with intent discovery and YAGNI review
- Evaluator-Optimizer Pattern - Automatic iteration cycles from Anthropic's agent architecture
- 9-Stage Development Pipeline - From schema design to deployment
- 3 Project Levels - Starter (static), Dynamic (fullstack), Enterprise (microservices)
- Multilingual Support - 8 languages (EN, KO, JA, ZH, ES, FR, DE, IT)
- 28 Skills - Domain-specific knowledge for various development scenarios
- 17 Agents - Specialized AI assistants including CTO-Led Team agents
- Check-Act Iteration Loop - Automatic gap analysis and fix cycles with max 5 iterations (90% threshold)
- bkend.ai BaaS Integration - Backend-as-a-Service skills for fullstack development
| Requirement | Minimum Version | Notes |
|---|---|---|
| OpenCode | 0.1+ | Required. bkit uses @opencode-ai/plugin ^1.2.4 |
| Bun | 1.3+ | OpenCode's runtime |
Add bkit-opencode to your project's opencode.json:
OpenCode will automatically install the plugin from npm on next launch.
# Clone for local development
git clone https://github.com/popup-studio-ai/bkit-opencode.gitThen add as a local plugin in your project's opencode.json:
{
"plugin": ["file:///path/to/bkit-opencode/src/index.ts"]
}bkit-opencode/
├── agents/ # 17 specialized AI agents
│ ├── cto-lead.md # Team orchestrator (opus tier)
│ ├── code-analyzer.md # Code quality analysis (opus tier)
│ ├── gap-detector.md # Design-implementation gap detection (opus tier)
│ ├── frontend-architect.md # UI/UX architecture (sonnet tier)
│ └── ... # 13 more agents
├── skills/ # 28 domain-specific skills
│ ├── pdca/SKILL.md # Unified PDCA with 8 actions
│ ├── plan-plus/SKILL.md # Enhanced planning with brainstorming
│ ├── starter/SKILL.md # Static web development
│ ├── dynamic/SKILL.md # Fullstack with BaaS
│ ├── enterprise/SKILL.md # Microservices architecture
│ └── ... # 23 more skills
├── src/ # TypeScript plugin source
│ ├── index.ts # Plugin entry point
│ ├── hooks/ # 8 hook handlers
│ ├── tools/ # 8 custom tools
│ └── lib/ # Core libraries
│ ├── core/ # Config, cache, platform, debug
│ ├── pdca/ # Status, level, phase, automation
│ ├── intent/ # Trigger detection, ambiguity, language
│ ├── task/ # Classification, tracking, delegation
│ └── team/ # Coordination, orchestration, mailbox
├── templates/ # 17 document templates
│ ├── plan.template.md
│ ├── design.template.md
│ ├── analysis.template.md
│ └── pipeline/ # 9-phase pipeline templates
├── bkit.config.json # Centralized configuration
└── package.json
/starter # Static website (Starter level)
/dynamic # Fullstack with BaaS (Dynamic level)
/enterprise # Microservices with K8s (Enterprise level)/pdca plan {feature} # Create plan document
/pdca design {feature} # Create design document
/pdca do {feature} # Implementation guide
/pdca analyze {feature} # Run gap analysis
/pdca iterate {feature} # Auto-fix with Evaluator-Optimizer pattern
/pdca report {feature} # Generate completion report
/pdca status # Check current PDCA status
/pdca next # Guide to next PDCA step# Start CTO Team for a feature
/pdca team {feature}How it works:
- CTO lead agent analyzes the feature and selects the optimal team composition
- Teammates are spawned in parallel (Dynamic: 3, Enterprise: 5 agents)
- Each teammate handles a specific area (QA, frontend, backend, security, etc.)
- CTO orchestrates task assignment, progress monitoring, and result aggregation
bkit uses a tier-based model system. Configure models in your project's bkit.config.json:
// bkit.config.json (in project root)
{
"models": {
"opus": { "providerID": "anthropic", "modelID": "claude-opus-4" },
"sonnet": { "providerID": "anthropic", "modelID": "claude-sonnet-4-5" },
"haiku": { "providerID": "anthropic", "modelID": "claude-haiku-4-5" }
}
}If no models are configured, agents use OpenCode's default model.
Agent Tier Mapping:
| Tier | Agents | Use Case |
|---|---|---|
| opus | cto-lead, gap-detector, code-analyzer, security-architect, enterprise-expert, infra-architect | Complex analysis, orchestration |
| sonnet | frontend-architect, backend-expert, baas-expert, product-manager, qa-strategist, design-validator, pdca-iterator, starter-guide, pipeline-guide | Implementation, review |
| haiku | qa-monitor, report-generator | Monitoring, reporting |
{
"pdca": {
"matchRateThreshold": 90, // Gap analysis pass threshold (%)
"autoIterate": true, // Auto-improvement when < threshold
"maxIterations": 5 // Max iteration cycles
}
}Comprehensive Guide: See CUSTOMIZATION-GUIDE.md for detailed instructions.
OpenCode searches for configuration in this priority order:
- User's
opencode.jsonagent/skill config (highest priority) - Project
.opencode/agents/and.opencode/skill/(user overrides) - Plugin defaults (bkit-opencode)
Create a file with the same name in .opencode/agents/:
# Copy and customize an agent
mkdir -p .opencode/agents
# Create .opencode/agents/cto-lead.md with your custom promptAdd skills to .opencode/skill/:
mkdir -p .opencode/skill/my-custom-skill
# Create .opencode/skill/my-custom-skill/SKILL.md| Level | Description | Stack | Detection |
|---|---|---|---|
| Starter | Static websites, portfolios | HTML, CSS, JS | Default |
| Dynamic | Fullstack applications | Next.js, BaaS | api/, supabase/, @bkend |
| Enterprise | Microservices architecture | K8s, Terraform, MSA | kubernetes/, terraform/, Dockerfile |
| Agent | Tier | Role |
|---|---|---|
| cto-lead | opus | Team orchestration, PDCA workflow management |
| code-analyzer | opus | Code quality, security, and performance analysis |
| gap-detector | opus | Design-implementation synchronization check |
| security-architect | opus | Vulnerability analysis, auth design review |
| enterprise-expert | opus | CTO-level AI-Native development strategy |
| infra-architect | opus | AWS, Kubernetes, Terraform infrastructure design |
| frontend-architect | sonnet | UI/UX design, component architecture |
| backend-expert | sonnet | Backend architecture across all frameworks |
| baas-expert | sonnet | bkend.ai BaaS platform integration |
| product-manager | sonnet | Requirements analysis, feature prioritization |
| qa-strategist | sonnet | Test strategy, quality metrics coordination |
| design-validator | sonnet | Design document completeness validation |
| pdca-iterator | sonnet | Evaluator-Optimizer automatic iteration |
| starter-guide | sonnet | Beginner-friendly step-by-step guidance |
| pipeline-guide | sonnet | 9-phase development pipeline guidance |
| qa-monitor | haiku | Docker log monitoring, Zero Script QA |
| report-generator | haiku | PDCA completion report generation |
pdca- Unified PDCA with 8 actions (plan, design, do, analyze, iterate, report, status, next)plan-plus- Brainstorming-enhanced planningbkit-rules- Core PDCA automation rulesbkit-templates- Document template reference
starter- Static web developmentdynamic- Fullstack with BaaSenterprise- Microservices architecture
phase-1-schemathroughphase-9-deploymentdevelopment-pipeline- Pipeline overview and navigation
bkend-quickstart- bkend.ai onboardingbkend-auth- Authentication and securitybkend-data- Database operationsbkend-storage- File storagebkend-cookbook- Practical tutorialsbackend-guide- Multi-framework backend reference
code-review- Code quality analysiszero-script-qa- Log-based testing without scriptsopencode-learning- OpenCode learning guidemobile-app- React Native / Flutterdesktop-app- Electron / Tauri
bkit automatically detects your language from trigger keywords:
| Language | Trigger Keywords |
|---|---|
| English | static website, beginner, API design |
| Korean | 정적 웹, 초보자, API 설계 |
| Japanese | 静的サイト, 初心者, API設計 |
| Chinese | 静态网站, 初学者, API设计 |
| Spanish | sitio web estatico, principiante |
| French | site web statique, debutant |
| German | statische Webseite, Anfanger |
| Italian | sito web statico, principiante |
- CUSTOMIZATION-GUIDE.md - Complete customization guide
- AI-NATIVE-DEVELOPMENT.md - AI-Native methodology
- CHANGELOG.md - Version history
- CONTRIBUTING.md - Contribution guidelines
We welcome contributions! Please see CONTRIBUTING.md for guidelines.
Copyright 2024-2026 POPUP STUDIO PTE. LTD.
Licensed under the Apache License, Version 2.0. See LICENSE for details.
You must include the NOTICE file in any redistribution.
- Issues: GitHub Issues
- Email: contact@popupstudio.ai
- Claude Code version: bkit-claude-code
Made with AI by POPUP STUDIO