Welcome to AIKit

Transform your AI coding assistant into a production-grade development environment.

AIKit is an open-source toolkit that supercharges Google Antigravity (Google DeepMind's AI IDE), OpenCode, Cursor, and Claude Code with structured workflows, visual diagramming, task tracking, and automated quality gates. Built for developers who want reliable, repeatable AI-powered development.

🚀 Featured: Google Antigravity Integration

AIKit now supports Google Antigravity - Google DeepMind's next-generation AI coding experience!

# Initialize with Antigravity
aikit init antigravity

# Install skills to .agent/skills/
aikit install antigravity

Why Antigravity + AIKit?

🧠 Gemini-powered - Leverage Google's most advanced AI
📁 Native skill format - Skills installed as SKILL.md files
🔄 Progressive disclosure - Agent discovers skills automatically
⚡ Zero config - Works out of the box

Get Started with Antigravity →

Hot Features

Interactive Diagrams with Draw.io

Create and edit diagrams visually - AI generates, you refine.

/drawio-interact create user-authentication
# AI generates diagram → opens in Draw.io → edit visually
/drawio-interact sync-to-mmd user-authentication
# Sync back to Mermaid code for version control

Why developers love it:

✅ Describe in plain English → AI generates diagrams instantly
✅ Edit visually with drag-drop (no Mermaid syntax needed)
✅ Bidirectional sync with manual control (reliable, not flaky)
✅ Version control with Mermaid files
✅ Works offline with Draw.io desktop app

Perfect for: Architecture diagrams, flowcharts, API design, system design, database schemas

Learn more →

Structured Workflows (22 Built-in Skills)

Stop leaving quality to chance. Enforce best practices automatically.

/use_skill "test-driven-development"
Implement user authentication
# Agent follows: RED → GREEN → REFACTOR

Popular Skills:

TDD - Automatic test-driven development workflow
Refactoring - Code cleanup with quality checks
Error Handling - Robust error management patterns
Code Review - Automated security and quality audits
API Design - RESTful API best practices
Component Design - React/Next.js component patterns

Result: Consistent code quality across your entire team.

Explore all 22 skills →

Task Tracking with Quality Gates

Ship features with confidence. Every task gets tracked, tested, and verified.

/create Add user authentication
/plan authentication system
/implement
/finish  # Runs: typecheck → test → lint → build

What /finish checks:

✅ TypeScript type checking passes
✅ All tests pass (unit + integration)
✅ Linting passes (ESLint/Prettier)
✅ Build succeeds
✅ No regressions introduced

Result: Zero surprises in production.

Learn about Beads →

Specialized AI Agents (8 Types)

Right agent for the right job. Automatic delegation for complex tasks.

@planner Design microservices architecture
# Delegates to @scout for research
# Delegates to @build for implementation
# Delegates to @review for verification

Available Agents:

@planner - Strategic planning & coordination
@build - Feature implementation
@rush - Quick fixes (hotfix mode)
@review - Code review & security audits
@scout - External research (docs, GitHub, npm)
@explore - Fast codebase navigation
@vision - Multimodal (images, PDFs, diagrams)
@one-shot - End-to-end autonomous execution

Meet the agents →

Session Continuity & Memory

Never lose context. Pick up exactly where you left off.

/handoff
# Saves: files changed, git state, progress notes, decisions
# ... time passes ...
/resume
# Loads everything back - continue seamlessly

What gets saved:

Modified files and git state
Progress notes and decisions made
Linked Beads task context
Branch and commit information
Next steps and blockers

Result: Seamless collaboration across days, weeks, or team members.

Learn about sessions →

What Makes AIKit Different

Before AIKit

❌ No structure - AI does whatever it wants
❌ No tracking - What did we work on yesterday?
❌ No quality gates - Bugs slip into production
❌ No continuity - Start from scratch every session
❌ No memory - Decisions get forgotten
❌ Manual checks - Did you run tests? Lint? Typecheck?

After AIKit

✅ Structured workflows - Enforced best practices
✅ Task tracking - Know exactly what's being worked on
✅ Quality gates - Automatic verification in /finish
✅ Session continuity - Resume across sessions
✅ Persistent memory - Handoffs capture everything
✅ One command - /finish runs all quality checks

Quick Wins in 5 Minutes

1. Ship Code with Quality Gates

/implement
# AI writes code with TDD workflow
/finish
# Automatically: typecheck → test → lint → build

Time saved: 15 minutes per feature (no manual checks)

2. Create Architecture Diagrams

/drawio-interact create microservices-architecture
# Edit in Draw.io, save
/drawio-interact sync-to-mmd microservices-architecture

Time saved: 30 minutes (no manual Mermaid syntax)

3. Resume Yesterday's Work

/session-current
# See what you were working on
/session-show <session-id>
# Load full context with /resume

Time saved: 10 minutes (no "what was I doing?" confusion)

By The Numbers

Metric	Without AIKit	With AIKit
Workflow Enforcement	Ad-hoc	22 structured skills
Quality Gates	Manual (forgotten 50% of time)	Automatic (100%)
Session Continuity	None	Full context handoffs
Task Tracking	Mental/external tools	Built-in Beads integration
Code Review Time	30+ minutes	5 minutes (automated)
Diagram Creation	60+ minutes	5 minutes (AI-generated)
Onboarding New Devs	2+ weeks	3 days (structured workflows)

What AIKit Solves

Problem 1: Inconsistent AI Output

Issue: Claude Code gives different answers each time, skips steps, forgets best practices.

Solution: 22 built-in skills enforce structured workflows (TDD, refactoring, error handling).

Problem 2: No Quality Enforcement

Issue: Bugs slip through because someone forgot to run tests/lint/typecheck.

Solution: /finish command runs automatic quality gates before shipping.

Problem 3: Lost Context Between Sessions

Issue: "What were we working on yesterday?" → Wasted time re-learning context.

Solution: Session handoffs capture everything - files, git state, decisions, next steps.

Problem 4: Manual Diagram Creation is Slow

Issue: Creating Mermaid diagrams requires learning syntax, takes forever.

Solution: Describe it → AI generates → Edit visually in Draw.io → Sync back to code.

Problem 5: No Task Visibility

Issue: "Is feature X done? What's blocking Y?" → No clear answer.

Solution: Beads task tracking shows status, blockers, and quality gate results.

Complete Feature List

22 Built-in Skills - TDD, refactoring, debugging, API design, component design, database design, security audits, and more
8 Specialized Agents - Planner, Builder, Rush, Reviewer, Scout, Explorer, Vision, One-Shot
50+ Slash Commands - Quick shortcuts for every workflow
Task Tracking - Beads integration with quality gates
Session Management - Track work across sessions with handoffs
Quality Gates - Automatic typecheck, test, lint, build in /finish
Anti-Hallucination - 3-layer system to prevent AI errors
Persistent Memory - Cross-session context with automatic handoffs
Draw.io Integration - Visual diagram editing with bidirectional Mermaid sync
Figma Integration - Extract design tokens and automate frontend development
MCP Server - Full Claude Code and OpenCode integration
Code Review Automation - Security audits, performance checks, best practice validation

Who Uses AIKit?

Frontend Developers

Component design patterns
State management workflows
Accessibility enforcement
Performance optimization

Backend Developers

API design best practices
Database schema design
Error handling patterns
Security audits

Full-Stack Developers

System architecture diagrams
TDD workflows
Code refactoring
Quality gates

Tech Leads & Architects

Architecture visualization
Team workflow standardization
Code review automation
Knowledge management

Get Started in 3 Steps

1. Install AIKit

npm install -g @tdsoft-tech/aikit

2. Initialize in Your Project

aikit init --global
cd your-project
aikit init
aikit install

3. Create Your First Task

/create Add user authentication

Or try interactive diagrams:

/drawio-interact create user-flow

Recommended Next Steps

New to AIKit?

Quick Start Guide - Get started in 5 minutes
Feature Overview - Understand all capabilities
Installation Guide - Detailed setup instructions

Want to Ship Better Code?

Quality Gates - Task tracking with /finish
TDD Workflow - Test-driven development
Code Review - Automated reviews

Interested in Diagrams?

Draw.io Integration - Visual diagram editing
System Design - Architecture design workflow

Want to Understand AIKit?

Skills Guide - All 22 built-in skills
Agents Guide - Meet the 8 specialized agents
Commands Reference - All 50+ commands

Why Developers Love AIKit

"AIKit transformed how we use Claude Code. No more inconsistent outputs - every feature follows the same TDD workflow, and /finish catches bugs before they reach production."

— Full-Stack Developer, Startup

"The Draw.io integration is a game-changer. I describe what I want, AI generates the diagram, I tweak it visually, and it syncs back to code. What used to take an hour now takes 5 minutes."

— Tech Lead, E-commerce Platform

"Session handoffs mean I never lose context. Whether it's been a day or a week, /resume brings me right back to where I left off."

— Backend Developer, Fintech Company

Join the Community

GitHub Repository - Star us on GitHub
Report Issues - Bug reports and feature requests
Discussions - Ask questions, share ideas
Documentation - Contribute to docs

Ready to Supercharge Claude Code?

npm install -g @tdsoft-tech/aikit
aikit init --global

Ship better code, faster, with confidence.

🚀 Featured: Google Antigravity Integration

auto_awesome Hot Features

brush Interactive Diagrams with Draw.io

track_changes Structured Workflows (22 Built-in Skills)

task_alt Task Tracking with Quality Gates

smart_toy Specialized AI Agents (8 Types)

storage Session Continuity & Memory

local_fire_department What Makes AIKit Different