Complete reference documentation for the Agentic Kit plugin system
Last Updated: November 2025 | Version: 1.1.0
- Overview
- Architecture
- Agents Reference
- Skills Reference
- Variants Comparison
- Installation & Configuration
- Advanced Usage
- Development Guide
- Troubleshooting
- FAQ
Agentic Kit is a comprehensive plugin for Claude Code that provides:
- 13 Specialized AI Agents for different development roles
- 14 Powerful Skills for document generation, design, and automation
- 3 Variants (Lite, Standard, Pro) for different use cases
- Persistent Session State - Skills and agents load automatically
- Auto-Discovery - Agents register and become available instantly
Agents are specialized AI assistants with specific roles and expertise:
- Invoked using
@AgentName: [your request] - Each agent has defined capabilities and workflows
- Agents can collaborate and hand off tasks
Skills are executable capabilities that agents can use:
- Document generation (PDF, DOCX, XLSX, PPTX)
- Design and branding tools
- Development utilities
- Testing frameworks
Variants control which agents and skills are available:
- Lite: Minimal footprint (3 agents, 0 skills)
- Standard: Balanced (13 agents, 8 skills) - Recommended
- Pro: Full feature set (13 agents, 14 skills)
Agentic Kit
├── .claude-plugin/
│ ├── plugin.json # Main manifest
│ ├── plugin-lite.json # Lite variant manifest
│ ├── plugin-standard.json # Standard variant manifest
│ ├── plugin-pro.json # Pro variant manifest
│ └── marketplace.json # Marketplace catalog
├── agents/ # 13 agent definitions
├── skills/ # 14 skill implementations
├── hooks/
│ ├── register-agents.js # Plugin-load hook
│ └── session-start.js # Session-start hook (NEW!)
└── resources/ # Templates, data, workflows
- Installation: User installs via Claude Code or npm
- Plugin Load:
register-agents.jsruns, registers all agents - Session Start:
session-start.jsruns every time Claude Code starts- Loads variant-specific skills
- Initializes agents
- Displays startup banner
- Runtime: Agents and skills available for use
- Session End: State persists for next session
Inspired by obra/superpowers, we now auto-load skills on every session start:
Benefits:
- ✅ Skills always available (no manual activation)
- ✅ Consistent experience across sessions
- ✅ Variant-specific features load automatically
- ✅ Startup feedback shows what's loaded
How it works:
// hooks/session-start.js
- Loads all skills defined in active variant manifest
- Registers agents for auto-invocation
- Displays banner showing agent/skill counts
- Returns session contextFile: agents/master.md
ID: master
Variants: All
Purpose: Central coordination and workflow orchestration
Capabilities:
- Route requests to appropriate specialized agents
- Coordinate multi-agent workflows
- Provide high-level guidance and help
- Manage complex multi-step tasks
Example Usage:
@Master: help
@Master: Create a product requirements document for a mobile app
@Master: Show me available agents
File: agents/orchestrator.md
ID: orchestrator
Variants: All
Purpose: Workflow management and multi-agent task coordination
Capabilities:
- Design multi-agent workflows
- Sequence complex operations
- Handle dependencies between tasks
- Monitor workflow progress
Example Usage:
@Orchestrator: Plan a workflow for building a new feature
@Orchestrator: Coordinate backend and frontend development
File: agents/story-writer.md
ID: story-writer
Variants: All
Purpose: Agile project management and team coordination
Capabilities:
- Create and manage user stories
- Plan sprints and iterations
- Facilitate retrospectives
- Track progress and blockers
Example Usage:
@ScrumMaster: Create user stories for the authentication feature
@ScrumMaster: Plan our next sprint
File: agents/feature-planner.md
ID: feature-planner
Variants: Standard, Pro
Purpose: Product strategy, vision, and roadmap planning
Capabilities:
- Create Product Requirements Documents (PRDs)
- Define product vision and strategy
- Prioritize features and epics
- Conduct market analysis
Example Usage:
@ProductManager: Create a PRD for a task management app
@ProductManager: Help me prioritize our backlog
File: agents/backlog-manager.md
ID: backlog-manager
Variants: Standard, Pro
Purpose: Backlog management and requirement refinement
Capabilities:
- Refine user stories
- Define acceptance criteria
- Manage product backlog
- Validate story completeness
Example Usage:
@ProductOwner: Refine this user story with acceptance criteria
@ProductOwner: Review our backlog for completeness
File: agents/market-researcher.md
ID: market-researcher
Variants: Standard, Pro
Purpose: Business requirements analysis and strategic insights
Capabilities:
- Analyze business requirements
- Conduct competitive research
- Document business processes
- Facilitate stakeholder interviews
Example Usage:
@BusinessAnalyst: Analyze the market for expense tracking apps
@BusinessAnalyst: Document our customer onboarding process
File: agents/code-developer.md
ID: code-developer
Variants: Standard, Pro
Purpose: Full-stack implementation and code development
Capabilities:
- Implement features (frontend + backend)
- Write tests and debug code
- Refactor and optimize
- Follow development best practices
Example Usage:
@FullStackDev: Implement the login API endpoint
@FullStackDev: Write tests for the authentication flow
File: agents/system-architect.md
ID: system-architect
Variants: Standard, Pro
Purpose: System architecture and technical design
Capabilities:
- Design system architecture
- Plan microservices
- Document technical decisions
- Evaluate technology stacks
Example Usage:
@HolisticArchitect: Design a scalable architecture for our platform
@HolisticArchitect: Recommend a tech stack for our API
File: agents/ui-designer.md
ID: ui-designer
Variants: Standard, Pro
Purpose: User experience and interface design
Capabilities:
- Create wireframes and prototypes
- Design user flows
- Conduct usability analysis
- Define interaction patterns
Example Usage:
@UXExpert: Create wireframes for the dashboard
@UXExpert: Design the user onboarding flow
File: agents/quality-assurance.md
ID: quality-assurance
Variants: Standard, Pro
Purpose: Quality assurance and testing strategy
Capabilities:
- Design test strategies
- Define test cases
- Review code quality
- Assess production readiness
Example Usage:
@QATestArchitect: Create a test plan for the payment feature
@QATestArchitect: Review this code for quality issues
File: agents/1-create-prd.md
ID: 1-create-prd
Variants: Standard, Pro
Purpose: Generate Product Requirements Documents
Capabilities:
- Structured PRD creation
- Requirements discovery
- Feature documentation
- Stakeholder alignment
Example Usage:
@CreatePRD: Generate a PRD for a mobile expense tracker
File: agents/2-generate-tasks.md
ID: 2-generate-tasks
Variants: Standard, Pro
Purpose: Break down features into implementation tasks
Capabilities:
- Convert PRDs to task lists
- Decompose epics into stories
- Estimate effort
- Sequence dependencies
Example Usage:
@GenerateTasks: Create tasks from this PRD
@GenerateTasks: Break down the authentication epic
File: agents/3-process-task-list.md
ID: 3-process-task-list
Variants: Standard, Pro
Purpose: Execute and track task lists systematically
Capabilities:
- Sequential task execution
- Progress tracking
- Commit management
- Test-first workflow
Example Usage:
@ProcessTaskList: Execute this task list
@ProcessTaskList: Continue from where we left off
Path: skills/pdf/
Variants: Standard, Pro
Capabilities:
- Generate PDF documents
- Manipulate existing PDFs
- Extract text and data
- Merge and split PDFs
Path: skills/docx/
Variants: Standard, Pro
Capabilities:
- Create Word documents
- Apply formatting and styles
- Insert tables and images
- Generate templates
Path: skills/xlsx/
Variants: Standard, Pro
Capabilities:
- Create spreadsheets
- Analyze data
- Generate charts
- Apply formulas
Path: skills/pptx/
Variants: Standard, Pro
Capabilities:
- Create presentations
- Design slides
- Insert media
- Apply themes
Path: skills/canvas-design/
Variants: Standard, Pro
Capabilities:
- Create visual designs
- Generate artwork
- Apply typography
- Export in multiple formats
Path: skills/theme-factory/
Variants: Standard, Pro
Capabilities:
- Create design themes
- Apply styling systems
- Generate color palettes
- Define component styles
Path: skills/brand-guidelines/
Variants: Standard, Pro
Capabilities:
- Define brand identity
- Apply brand compliance
- Create style guides
- Maintain consistency
Path: skills/internal-comms/
Variants: Standard, Pro
Capabilities:
- Create internal documents
- Generate communication templates
- Format company materials
- Apply company branding
Path: skills/algorithmic-art/
Variants: Pro
Capabilities:
- Generate generative art
- Create algorithmic designs
- Produce procedural graphics
- Export artwork
Path: skills/artifacts-builder/
Variants: Pro
Capabilities:
- Build complex HTML artifacts
- Create interactive components
- Generate ShadCN components
- Bundle for deployment
Path: skills/mcp-builder/
Variants: Pro
Capabilities:
- Create Model Context Protocol servers
- Define MCP tools and resources
- Build custom integrations
- Package MCP servers
Path: skills/skill-creator/
Variants: Pro
Capabilities:
- Create custom skills
- Define skill interfaces
- Document skill usage
- Package for distribution
Path: skills/webapp-testing/
Variants: Pro
Capabilities:
- Test web applications
- Run automated tests
- Validate functionality
- Generate test reports
Path: skills/slack-gif-creator/
Variants: Pro
Capabilities:
- Create animated GIFs
- Optimize for Slack
- Apply branding
- Export in optimal format
| Feature | Lite | Standard | Pro |
|---|---|---|---|
| Agents | 3 | 13 | 13 |
| Skills | 0 | 8 | 14 |
| Document Generation | ❌ | ✅ | ✅ |
| Design Tools | ❌ | ✅ | ✅ |
| Advanced Skills | ❌ | ❌ | ✅ |
| Product Management | ❌ | ✅ | ✅ |
| Development Tools | ❌ | ✅ | ✅ |
| Size (unpacked) | ~2 MB | ~7 MB | ~9.5 MB |
| Installation Time | <1 min | <3 min | <5 min |
Choose Lite if:
- You only need basic agent orchestration
- Working in resource-constrained environments
- Building CI/CD pipelines
- Want minimal dependencies
Choose Standard if: ⭐ (Recommended)
- General-purpose development
- Need document generation (PDF, DOCX, etc.)
- Product management workflows
- Content creation
Choose Pro if:
- Advanced design work
- Building custom skills/tools
- Web app testing
- Maximum capabilities needed
See README.md for quick installation
Create your own marketplace catalog:
// .claude-plugin/marketplace.json
{
"name": "My Custom Marketplace",
"plugins": [
{
"name": "liteagents-standard",
"source": "github:hamr0/liteagents"
}
]
}Clone and link for development:
git clone https://github.com/hamr0/liteagents.git
cd liteagents
npm linkVariants are configured in manifest files:
.claude-plugin/plugin-lite.json.claude-plugin/plugin-standard.json.claude-plugin/plugin-pro.json
Chain agents for complex tasks:
# Step 1: Create requirements
@ProductManager: Create a PRD for user authentication
# Step 2: Generate tasks
@GenerateTasks: Break down this PRD into implementation tasks
# Step 3: Execute
@FullStackDev: Implement the first task
# Step 4: Test
@QATestArchitect: Create tests for this implementation
- Create
agents/my-agent.md - Add to variant manifest:
{
"id": "my-agent",
"name": "My Agent",
"description": "Custom agent description",
"path": "agents/my-agent.md"
}- Restart session
- Create
skills/my-skill/SKILL.md - Implement skill logic
- Add to variant manifest
- Restart session
Problem: Skills don't activate on session start
Solution:
- Check
hooks/session-start.jsis in manifest - Verify skill paths in variant manifest
- Restart Claude Code session
- Check console for errors
Problem: @AgentName: doesn't work
Solution:
- Verify agent is in your variant (check manifest)
- Check agent file exists in
agents/directory - Ensure
register-agents.jshook ran - Try reloading plugin
Problem: Wrong skills/agents available
Solution:
- Check which variant you installed
- Reinstall with correct variant
- Verify manifest file being used
Enable verbose logging:
# In session-start.js, set:
DEBUG=trueQ: Can I use multiple variants simultaneously? A: No, only one variant is active per installation. Reinstall to switch variants.
Q: Do skills persist across sessions?
A: Yes! The session-start hook loads skills automatically.
Q: Can I create custom agents?
A: Yes! Add markdown files to agents/ and update the manifest.
Q: How do I update to the latest version?
A: npm update liteagents or reinstall via Claude Code.
Q: Is there a Pro trial? A: All variants are free and open-source. Install Pro to try it.
Q: Can agents collaborate? A: Yes! Agents can hand off tasks and work together.
Q: What's the difference between skills and agents? A: Agents are AI assistants with roles. Skills are executable tools agents can use.
- GitHub: https://github.com/hamr0/liteagents
- npm: https://www.npmjs.com/package/liteagents
- Issues: https://github.com/hamr0/liteagents/issues
- Detailed Docs:
- AGENTS.md - Complete agent directory
- SKILLS.md - Complete skill reference
- ARCHITECTURE.md - Technical architecture
- CONTRIBUTING.md - Contribution guidelines
- TROUBLESHOOTING.md - Detailed troubleshooting
- PUBLISHING.md - Publishing guide
Last Updated: November 2025 Version: 1.1.0 Maintainer: hamr0