Skip to main content

AI-native personal operations system — coordinated agents for your business

Project description

RealizeOS

RealizeOS

The AI operations system for your business.
Coordinated AI agents that understand your venture, remember your preferences,
and execute multi-step workflows — not just another chatbot.

CI Release PyPI npm Docker License: BSL 1.1 Python 3.11+ Stars

⚡ Quickstart · 🏗️ Architecture · ✨ Features · 🔌 MCP Server · 💻 CLI Reference · 🚀 Self-Host · 🤝 Contribute


What is RealizeOS?

RealizeOS is a self-hosted AI operations system that gives your business a coordinated team of AI agents. Unlike generic chatbots, RealizeOS agents know your venture, run multi-step workflows, route to the optimal model, and respect governance with approval gates and audit logs.


⚡ Quick Start

🐳 Docker (recommended)
git clone https://github.com/SufZen/RealizeOS-5.git && cd RealizeOS-5
cp .env.example .env       # Add your API key(s)
docker compose up --build   # Dashboard → http://localhost:8080

Or standalone:

docker run -d -p 8080:8080 -v realizeos-data:/app/data ghcr.io/sufzen/realizeos-5:latest
🐍 pip (no Docker)
pip install realize-os
realize-os init --template consulting
realize-os serve
# Dashboard → http://localhost:8080

Requires Python 3.11+. Works on Windows, macOS, and Linux.

📦 NPX
npx @realize-os/cli init my-business
cd my-business && npx @realize-os/cli start

Requires Node.js 18+ and Docker.

🖥️ One-Liner Install

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/SufZen/RealizeOS-5/main/scripts/install.sh | bash

Windows (PowerShell):

irm https://raw.githubusercontent.com/SufZen/RealizeOS-5/main/scripts/install.ps1 | iex
Method Requires Best For
Docker Docker Production, isolated deployments
pip Python 3.11+ Local dev, no Docker needed
NPX Node.js 18+ + Docker Fastest project scaffolding
One-liner bash / PowerShell Server deployment, CI/CD

📖 Full setup guide: QUICKSTART.md · Production: Self-Hosting Guide


✨ Key Features

🧠 FABRIC Knowledge System

Every venture's AI knowledge is organized into six layers:

Layer Purpose
Foundations Venture identity, voice, core standards
Agents AI team definitions and routing guide
Brain Domain knowledge, market data, expertise
Routines Skills, workflows, state maps, SOPs
Insights Memory: learning log, feedback, decisions
Creations Output: deliverables, drafts, final assets

🤖 Multi-LLM Routing

The engine classifies every task and selects the optimal model:

Task Type Model Examples
Simple Gemini Flash Status checks, formatting, lookups
Content Claude Sonnet Writing, analysis, summarization
Complex Claude Opus Strategy, multi-step reasoning

Providers auto-discovered at startup. Supports Claude, Gemini, OpenAI, and Ollama (local).

🔧 Agent System

  • Composable agents with scope, inputs, outputs, guardrails, and tools
  • Pipelines — sequential execution with Dev-QA retry loops
  • 7 handoff types — standard, QA-pass/fail, escalation, phase-gate, sprint, incident
  • Hot-reload — filesystem-watched agent registry
  • Per-agent SOUL — persistent identity: role, personality, expertise, communication style
  • Tool gating — per-agent allowlists/denylists

🧩 Extension System

Type Purpose Example
tool New capabilities Stripe, Twilio, custom APIs
channel Communication Slack, Discord, WhatsApp
integration Backend sync CRM, analytics
hook Event reactions Notifications, logging

🛠️ Tool Ecosystem

Category Capabilities
Google Workspace Gmail (8 tools), Calendar (4), Drive (9), Sheets (3)
Financial Stripe charges, subscriptions, invoices
Web Search (Brave API), page scraping, headless browser
MCP Connect to any MCP-compatible tool server
Messaging Agent-to-agent bus, human notifications, channels
Governance Human-in-the-loop approval workflows

📋 Business Templates

Pre-built venture configurations:

consulting · agency · portfolio · saas · ecommerce · accounting · coaching · freelance

realize-os init --template consulting

🔌 MCP Server

RealizeOS ships a built-in MCP server so any MCP-speaking agent can use it as a second brain:

  • 24 tools across 4 families: Chat & Status, KB Read, Ops, Admin
  • HTTP+SSE transport — works with Claude Desktop, Cursor, n8n, cloud routines
  • Same auth — Bearer JWT or API key, same roles and audit logs
  • Gated access — KB, ops, and admin tools independently toggleable
realize-os mcp serve --port 8080        # Start API + MCP together
realize-os mcp token --user owner       # Issue a bearer token

📖 Full details: docs/mcp-server.md


💻 Operator CLI

The realize-os CLI is a first-class operator interface:

realize-os serve                                # Start API + dashboard
realize-os chat "What's the pipeline status?"   # Quick query
realize-os repl --system realization-il          # Interactive REPL
realize-os kb search "investment thesis"         # Search knowledge base
realize-os config profile add prod --endpoint https://my-vps:8080

Both realize-os (pip-installed) and python cli.py (source checkout) work identically.

📖 Full command reference: docs/cli-reference.md


🛡️ Security & Governance

  • 5-layer security middleware: Security headers → Audit → Rate limiting → Injection guard → JWT auth
  • RBAC with 6 roles: owner, admin, operator, user, viewer, guest
  • Prompt injection scanner — pattern + heuristic + Unicode normalization
  • Human-in-the-loop approval gates for consequential actions
  • Audit logging — JSONL persistent logs with SSE streaming
  • Secret redaction in error responses and logs

📖 Documentation

Guide Description
⚡ Quickstart Zero to running in 10 minutes
🏗️ Architecture FABRIC, message flow, modules
💻 CLI Reference Full operator CLI command tree
🔌 MCP Server Built-in MCP server: tools, security, integration
📖 Getting Started First steps after setup
🔧 Configuration Customize your deployment
🚀 Self-Hosting Production deployment
✍️ Skill Authoring Create custom skills
📡 API Reference REST + MCP API documentation
📋 Upgrade from 5.0 Migration guide: 5.0.x → 5.1.0
🤝 Contributing Developer guide

Requirements

  • Python 3.11+ (3.12+ recommended)
  • At least one LLM API key (Anthropic, Google, OpenAI, or Ollama)
  • Docker 24.0+ (optional, for containerized deployment)
  • Node.js 20+ (optional, for dashboard development)

Community


License

RealizeOS is licensed under the Business Source License 1.1.

  • ✅ Free to use, modify, and self-host
  • ✅ Free for internal business operations
  • ✅ Converts to Apache 2.0 on March 26, 2030
  • ❌ Cannot offer as a hosted/managed service without a commercial license

For commercial licensing inquiries, contact realizeos@realization.co.il.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

realize_os-5.2.1.tar.gz (618.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

realize_os-5.2.1-py3-none-any.whl (639.1 kB view details)

Uploaded Python 3

File details

Details for the file realize_os-5.2.1.tar.gz.

File metadata

  • Download URL: realize_os-5.2.1.tar.gz
  • Upload date:
  • Size: 618.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for realize_os-5.2.1.tar.gz
Algorithm Hash digest
SHA256 5f8d7baca7fbc39c4e0a86585a00cff624c08160096443460c1736037cd22b9d
MD5 4b1909bf76ed187705e40c2e76d804fe
BLAKE2b-256 1d78444f03becb27f970b642ca9847d943fa502824d6564c0bd88047dc505866

See more details on using hashes here.

Provenance

The following attestation bundles were made for realize_os-5.2.1.tar.gz:

Publisher: release.yml on SufZen/RealizeOS-5

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file realize_os-5.2.1-py3-none-any.whl.

File metadata

  • Download URL: realize_os-5.2.1-py3-none-any.whl
  • Upload date:
  • Size: 639.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for realize_os-5.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 62afbf6f6724f14db61eb261cdf3d9863a4d3e40462a312d9ec28b0945ad8ca4
MD5 f144b325cf438d8ddc5916f25d1499f7
BLAKE2b-256 268815d1365c632dd1404eda183fb1690f2241a599e3273c3fc79481236ca420

See more details on using hashes here.

Provenance

The following attestation bundles were made for realize_os-5.2.1-py3-none-any.whl:

Publisher: release.yml on SufZen/RealizeOS-5

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page