Claude Code Agents

Specialized AI agents for Claude Code — organized by domain. Multi-plugin marketplace.

Repository Structure

claude-code-agents/
├── .claude-plugin/
│   └── marketplace.json          # Marketplace catalog (2 plugins)
├── plugins/
│   ├── ai-sdlc/                  # Plugin: AI-powered SDLC agents
│   │   ├── .claude-plugin/
│   │   │   └── plugin.json
│   │   └── agents/
│   └── ai-university/            # Plugin: AI University
│       ├── .claude-plugin/
│       │   └── plugin.json
│       ├── skills/study/SKILL.md # /ai-university:study skill
│       ├── agents/education/
│       ├── config/
│       └── scripts/
├── agents/                       # Original agent files (for direct use)
│   ├── coding/kotlin/
│   └── education/
├── config/
│   └── university.yaml
├── scripts/
│   └── generate-harness.sh
└── README.md

Installation

Option 1: Install from GitHub (Recommended)

# Add the marketplace
/plugin marketplace add https://github.com/AlexGladkov/claude-code-agents

# Install individual plugins
/plugin install ai-sdlc
/plugin install ai-university

Option 2: Install from Local Clone

# Clone the repository
git clone https://github.com/AlexGladkov/claude-code-agents.git

# Add the local marketplace
/plugin marketplace add ./claude-code-agents

# Install individual plugins
/plugin install ai-sdlc
/plugin install ai-university

Available Plugins

Plugin	Description
ai-sdlc	AI-powered SDLC agents for Kotlin/Spring Boot + Compose Multiplatform
ai-university	AI University — teaching agents for medicine, AI/ML

Available Agents

Coding / Kotlin

Located in agents/coding/kotlin/

Agent	Description
init-kotlin	Repository bootstrap for clean Kotlin Spring Boot or full-stack (Spring + Compose) projects
builder-spring-feature	Feature generation for Spring Boot with strict architecture (feature-slice, layering, dependency validation)
builder-compose-feature	Feature generation for Compose Multiplatform with Screen/View/Component separation and MVVM
test-spring	High-quality test automation following SDET/AQA practices, AAA pattern, Testcontainers integration
kotlin-diagnostics	Bug detection and diagnosis for Kotlin/Compose/Android/Spring with automatic runtime analysis
refactor-spring	Architectural refactoring of Spring applications enforcing SOLID, layering, file structure
refactor-mobile	Architectural refactoring of Android code (Clean Architecture, Compose, Decompose, Kodein)
security-kotlin	OWASP security auditing for Spring Boot with comprehensive vulnerability scanning
devops-orchestrator	Docker environment setup, multi-env configs, CI/CD pipelines, automated deployments
system-analytics	Technical specification generation from user requests, saved as structured Markdown
kotlin-multiplatform-developer	Full KMP feature slice generator (domain + data + presentation) with feature-sliced architecture

Education / University

The education agents are organized as a virtual university with a 4-level hierarchy:

University → Faculty → Department → Discipline (agent)

The university structure is defined in config/university.yaml (source-of-truth).

Quick start:

# Generate instruction files for your AI tool
./scripts/generate-harness.sh

# Generate and install (e.g., for Claude Code)
./scripts/generate-harness.sh --install --only claude

Supported AI tools: Claude Code, Cursor, GitHub Copilot, Windsurf, Codex/OpenCode.

Current structure:

Faculty	Department	Disciplines	Hours
Medical (medicine)	Biology	Anatomy, Physiology, Neurobiology	360
AI/ML (ai-ml)	AI SDLC	Prompting, MCP, RAG, Local AI, AI Security, CV, AI Agents, MLOps	560
AI/ML (ai-ml)	ML Foundations	Classical ML, Deep Learning, Transformers, Fuzzy Logic, Optimization, Generative Models, RL, RNN/TimeSeries, Graph NN	700
Robotics (robotics)	Robot Fundamentals	Kinematics & Dynamics, Sensors & Actuators, Control Systems, Electronics & MCU, Robot Math	360
Robotics (robotics)	Applied Robotics	ROS 2, Robot Vision, Motion Planning, SLAM, Simulation & Digital Twins	360

Key agents:

Agent	Description
rector	University rector — top-level orchestrator, study plans, cross-faculty programs, progress aggregation

Medical / Biology:

Agent	Description
department-head	Department head — learning orchestrator, prerequisites, interdisciplinary checks
anatomy-teacher	Human anatomy — systemic, clinical, topographic
physiology-teacher	Human physiology — all organ systems
neurobiology-teacher	Neurobiology — cellular and systems

AI/ML / AI SDLC:

Agent	Description
ai-sdlc-department-head	Department head — AI SDLC learning orchestrator
prompting-teacher	Prompt engineering, state & context management
mcp-teacher	Model Context Protocol — servers, tools, transports
rag-teacher	RAG & Embeddings — retrieval, vector DB, chunking
local-ai-teacher	Local AI — Ollama, llama.cpp, vLLM, quantization
ai-security-teacher	AI Security — OWASP LLM Top 10, prompt injection, red teaming
cv-teacher	Computer Vision — CNN, detection, segmentation, VLM
ai-agents-teacher	AI Agents — tool use, ReAct, multi-agent, memory
mlops-teacher	MLOps — serving, monitoring, pipelines, CI/CD

AI/ML / ML Foundations:

Agent	Description
ml-foundations-department-head	Department head — ML Foundations learning orchestrator
classical-ml-teacher	Classical ML — SVM, trees, ensembles, clustering, feature engineering
deep-learning-teacher	Deep Learning — backprop, CNN, optimizers, transfer learning
transformers-teacher	Encoder/Decoder & Transformers — attention, BERT/GPT/T5, tokenization
fuzzy-logic-teacher	Fuzzy Logic — fuzzy sets, Mamdani/Sugeno, ANFIS, neuro-fuzzy
optimization-teacher	Optimization & Learning Theory — convex/non-convex, PAC-learning, VC-dimension
generative-models-teacher	Generative Models — VAE, GAN, Diffusion, Flow-based
reinforcement-learning-teacher	Reinforcement Learning — MDP, DQN, PPO, RLHF, multi-agent
rnn-timeseries-teacher	RNN & Time Series — LSTM, GRU, TFT, forecasting, anomaly detection
graph-nn-teacher	Graph Neural Networks — GCN, GAT, knowledge graphs, molecular ML

Robotics / Fundamentals:

Agent	Description
robot-fundamentals-department-head	Department head — robotics fundamentals orchestrator
kinematics-dynamics-teacher	Kinematics & dynamics — DH, Jacobians, mobile robots
sensors-actuators-teacher	Sensors & actuators — LiDAR, IMU, motors, sensor fusion
control-systems-teacher	Control systems — PID, LQR, adaptive, robust, real-time
electronics-mcu-teacher	Electronics & MCU — STM32, ESP32, protocols, PCB, embedded Linux
robot-math-teacher	Math for robotics — SO(3)/SE(3), quaternions, Bayesian, optimization

Robotics / Applied:

Agent	Description
applied-robotics-department-head	Department head — applied robotics orchestrator
ros2-teacher	ROS 2 — DDS, Nav2, MoveIt 2, tf2, lifecycle
robot-vision-teacher	Robot vision — stereo, VO, 3D reconstruction, segmentation
motion-planning-teacher	Motion planning — RRT, A*, MoveIt, behavior trees, multi-robot
slam-teacher	SLAM — LiDAR/visual/multi-sensor SLAM, graph optimization
simulation-teacher	Simulation — Gazebo, Isaac Sim, MuJoCo, sim-to-real, digital twins

Development Workflow (Kotlin)

These agents cover the complete development lifecycle:

1. init-kotlin           --> scaffold new projects
        |
        v
2. builder-spring        --> generate backend features
   builder-compose       --> generate mobile features
        |
        v
3. test-spring           --> write comprehensive tests
        |
        v
4. diagnostics-kotlin    --> find and fix bugs
        |
        v
5. refactor-spring       --> clean up backend architecture
   refactor-mobile       --> clean up mobile architecture
        |
        v
6. security-kotlin       --> OWASP audit
        |
        v
7. devops-orchestrator   --> containerization and CI/CD

Architecture Patterns Enforced

Spring Boot (Backend)

• Feature-slice organization: feature/<name>/api/service/persistence/domain/
• Layered architecture: Controller -> Service -> Repository (no shortcuts)
• One type per file, max 1000 lines per file, max 100 lines per method

Compose Multiplatform (Frontend)

• Feature-slice: feature/<name>/screen/view/component/domain/data/di/
• MVVM with Decompose: Component holds state, View is pure UI
• Use cases always return Result<T>
• No remember() in Views, max 600 lines ideal per file

Common Rules

• Unidirectional data flow, no cyclic dependencies
• SOLID principles strictly enforced
• Tests follow AAA pattern with Testcontainers for external deps

Contributing

To suggest a new subagent, open a Pull Request with a markdown file in the appropriate category folder under agents/.

License

MIT