AutomatosX

Give your AI assistant a long-term memory and a team of autonomous agents.

AutomatosX is a CLI-first orchestration tool that transforms stateless AI assistants into a powerful, collaborative workforce. It provides persistent memory, intelligent agent delegation, and cross-provider support (Claude, Gemini, OpenAI), all running 100% locally.

Status: ✅ Production Ready · v5.8.2 · October 2025 · 19 Specialized Agents · 100% Resource Leak Free · Spec-Driven Development

Latest (v5.8.0): Spec-Kit Integration - AutomatosX now supports spec-driven development! Define your project specs in .specify/ directory (spec.md, plan.md, tasks.md) and let AutomatosX automatically orchestrate tasks based on your dependency graph. Features include: DAG-based task execution, cycle detection, LRU caching, and automatic spec detection. Perfect for structured, multi-step projects. See full changelog →

🚀 Get Started in 3 Steps

1. Install AutomatosX:

   npm install -g @defai.digital/automatosx

2. Initialize Your Project:

   cd your-project-folder
   ax init

3. Run Your First Agent:

   # Use natural language in Claude Code
   "please work with ax agent to design a REST API for user management"
   
   # Or run directly in your terminal
   ax run product "Design a REST API for user management"

➡️ Full Quick Start Guide | Installation Details

💡 Why AutomatosX? The End of AI Amnesia

The Problem: Stateless AI is Inefficient

Standard AI assistants suffer from digital amnesia. They have no memory of past conversations, forcing you to repeat context, re-explain requirements, and manually coordinate every single task. This is slow, repetitive, and inefficient.

• ❌ No Long-Term Memory: Every session starts from zero.
• ❌ Constant Repetition: You explain the same architecture and requirements over and over.
• ❌ Manual Coordination: You are the single point of failure for orchestrating tasks.
• ❌ Knowledge is Lost: Valuable insights and decisions disappear forever.

The Solution: A Persistent, Collaborative AI Workforce

AutomatosX gives your AI a permanent brain and a team of specialists who learn and collaborate.

• ✅ Persistent Memory: Agents remember everything. A local, zero-cost SQLite FTS5 database enables sub-millisecond context retrieval.
• ✅ Autonomous Delegation: Agents intelligently delegate tasks to each other, creating workflows that run on their own.
• ✅ Context on Autopilot: Agents automatically get the context they need from past conversations. Never repeat yourself again.
• ✅ A Team That Learns: Your AI team gets smarter and more effective with every task, building a shared knowledge base over time.

The result? You save hours per week, produce higher-quality work, and build a system that grows more valuable with every interaction.

🧠 Core Feature: Persistent Memory

AutomatosX remembers every conversation, decision, and piece of code automatically. This knowledge is instantly searchable and injected into future tasks, ensuring perfect context every time.

How It Works

# 1. A task is completed and automatically saved to memory.
ax run product "Design a calculator with add/subtract features"
# → Task and response are indexed in the local SQLite FTS5 database.

# 2. A related task is run later.
ax run backend "Implement the calculator"
# → Memory automatically finds the "calculator" design spec from the previous step.
# → The backend agent receives the design and can start work immediately.

• Technology: SQLite with FTS5 for fast, local full-text search.
• Speed: < 1ms search, even with thousands of entries.
• Cost: $0. No embedding APIs, no cloud services.
• Privacy: 100% local. Your data never leaves your machine.

➡️ Learn More: Memory System Guide

🤝 Core Feature: Multi-Agent Orchestration

Stop micromanaging. With AutomatosX, you can give a high-level goal to one agent, and it will create a plan, delegate tasks, and orchestrate a team of specialists to get the job done.

How It Works

// 1. You give a high-level task to a coordinator agent.
ax run product "Build a complete user authentication feature"

// 2. The Product agent creates a plan and delegates to other agents.
/*
  "I'll design the auth system with JWT and OAuth2.

   @backend Please implement the JWT authentication API based on this design.
   @security Please audit the implementation for security vulnerabilities."
*/

// 3. AutomatosX executes the plan automatically.
//    - The backend agent receives the spec and implements the API.
//    - The security agent receives the spec and the code, then performs an audit.
//    - Results are aggregated and returned.

• Natural Language Delegation: Use simple @mention syntax to delegate.
• Autonomous Workflows: Agents work in parallel to complete goals faster.
• Full Transparency: The entire delegation chain is tracked and logged.

➡️ Learn More: Multi-Agent Orchestration Guide

📋 Core Feature: Spec-Driven Development (✨ NEW in v5.8.0)

Define your project structure once, and let AutomatosX orchestrate execution automatically. Perfect for complex, multi-step projects with dependencies.

How It Works

# 1. Initialize spec-kit in your project
ax init --spec-kit

# 2. Define your project in .specify/ directory
.specify/
├── spec.md    # Requirements and success criteria
├── plan.md    # Technical approach and architecture
└── tasks.md   # Task breakdown with dependencies

# 3. Example tasks.md format:
- [ ] id:setup:env ops:"ax run backend 'Setup environment'"
- [ ] id:impl:api ops:"ax run backend 'Implement API'" dep:setup:env
- [ ] id:test:api ops:"ax run quality 'Test API'" dep:impl:api

# 4. AutomatosX automatically:
#    - Detects the spec in your workspace
#    - Builds a dependency graph (DAG)
#    - Executes tasks in the correct order
#    - Caches spec for fast access

Key Features:

• Automatic Spec Detection: Just create .specify/ and AutomatosX finds it
• DAG-Based Execution: Tasks run in dependency order with cycle detection
• LRU Caching: Fast spec loading with checksum-based invalidation
• Multi-Format Support: Flexible task format with 3 parsing patterns
• File Watching: Auto-reload when specs change (optional)

🎭 19 Specialized Agents

AutomatosX comes with a pre-built team of 19 agents, each with a specific role and expertise. This ensures the right specialist is always available for the task at hand.

• Astrid - Aerospace Scientist (Orbital mechanics, mission analysis, telemetry)
• Bob - Backend Engineer (API design, databases, Go/Rust systems)
• Candy - Creative Marketer (Content strategy)
• Daisy - Data Engineer (ETL, SQL, modeling)
• Dana - Data Scientist (ML strategy, statistical analysis)
• Debbee - UX Designer (UX research, wireframes)
• Eric - CEO (Business leadership)
• Felix - Fullstack Engineer (Node.js/TypeScript + Python)
• Frank - Frontend Engineer (React/Next.js/Swift UI)
• Maya - Mobile Engineer (iOS/Android, Swift/Kotlin/Flutter)
• Oliver - DevOps Engineer (CI/CD, infrastructure)
• Paris - Product Manager (Roadmaps, strategy)
• Peter - Best Practices Expert (SOLID, design patterns, clean code, refactoring, architecture) ✨ NEW in v5.6.21
• Queenie - Quality Specialist (Testing, code reviews)
• Quinn - Quantum Engineer (Quantum algorithms, Qiskit/Cirq, error correction)
• Rodman - Researcher (Feasibility studies)
• Steve - Security Engineer (Audits, threat modeling)
• Tony - CTO (Technical strategy)
• Wendy - Technical Writer (Documentation, reports)

Governance: Agents have clear roles, permissions, and delegation limits (maxDelegationDepth) to ensure efficient and safe collaboration while preventing infinite loops.

➡️ See the Full Agent Directory

🚀 Two Ways to Use AutomatosX

1. Claude Code Integration (Recommended)

Use natural language to collaborate with agents directly within your editor. Claude Code acts as an intelligent coordinator, providing project context and validating results.

# Let Claude Code think, plan, and coordinate
"please work with ax agent to implement user authentication"
"please work with ax agent to refactor this module with best practices"

For simple, direct tasks, use slash commands: /ax-agent backend, write a function to validate emails.

2. Terminal/CLI Mode (For Power Users)

Run agents directly from your terminal for scripting, automation, and CI/CD pipelines.

# Run a multi-agent workflow from your command line
ax run product "Design REST API for users"
ax run backend "Implement the API"           # Auto-receives design from memory
ax run quality "Write tests for the API" # Auto-receives design + implementation

➡️ Read the Terminal Mode Guide

🔄 Intelligent Provider Management (v5.7.0)

AutomatosX automatically handles provider usage limits so you never experience downtime.

Automatic Limit Detection & Rotation

When you hit a provider's usage limit, AutomatosX instantly detects it and switches to the next available provider:

$ ax run backend "implement authentication API"
⚠️  Switched from openai → gemini-cli
   (OpenAI daily quota hit, resets at 2025-10-29 00:00 UTC)

✓ Task completed successfully with gemini-cli

Auto-Recovery

The system tracks reset times (daily/weekly) and automatically restores providers when limits expire. No manual intervention needed.

Monitor Status

Check your provider status anytime:

$ ax provider-limits
📊 Provider Limits Status

  ⚠️  openai:
     Status: limited
     Window: daily
     Resets: 2025-10-29 00:00:00 (6h)

  ✅ 2 provider(s) available

Performance: < 1ms overhead per request, O(1) limit checks, persistent state across restarts.

➡️ Learn More: Provider Management Guide

✨ Key Features

📚 Documentation

• Quick Start Guide
• Core Concepts
• Full CLI Command Reference
• Agent Directory
• Workspace Path Conventions
• Troubleshooting Guide

🤝 Contributing

AutomatosX is an open-source project. We welcome contributions!

• Contributing Guide
• Report an Issue
• Request a Feature

📄 License

AutomatosX is Apache 2.0 licensed.