Mega-OS

A personal operating system for builders, powered by Claude Code. Mega-OS organizes 39 specialized AI agents into 5 categories, giving you a structured command center for managing products, business operations, and knowledge across every session.

What You Get

• 39 AI agents in 5 categories (Governance, Knowledge, Technical, Business, Evolution) with defined roles and handoff protocols
• 30 slash commands for daily operations, building, system management, and deployment
• 8 automated workflows (Planning, Technical, Business, Incident, Knowledge, Content, Site Build, Evolution)
• Optional cron automations for daily scans, news briefings, and system self-improvement
• A living task board in active/ that persists your context across every session
• Git-as-cloud-backup — every commit preserves your entire OS state

Getting Started

First install

git clone https://github.com/sly-the-fox/mega-os-public.git mega-os
cd mega-os
claude

Claude Code auto-loads the system. On first run, it detects an unconfigured system and prompts you to run /setup for guided interactive onboarding.

The /setup wizard

/setup           # Full guided setup (~20 min, 11 phases)
/setup --minimal # Quick setup (~5 min, gets you running fast)

Setup walks you through: environment verification, personalization, workspace scaffolding, standards customization, git cloud backup, Telegram bridge (optional), cron automations (optional), product scaffolding, and agent customization.

Cloud backup (recommended)

Mega-OS uses a two-remote model so you can receive framework updates while keeping your personal data private:

# Inside your mega-os directory after /setup, or manually:
git remote rename origin upstream
git remote add origin git@github.com:YOUR_USERNAME/my-mega-os.git  # private repo
git push -u origin master

• upstream — public Mega-OS repo (framework updates via /update)
• origin — your private repo (cloud backup of everything including personal data)

Every commit is a snapshot of your entire operating system. Push regularly and your OS is backed up, versioned, and accessible from any machine.

Second computer

git clone git@github.com:YOUR_USERNAME/my-mega-os.git mega-os
cd mega-os
git remote add upstream https://github.com/sly-the-fox/mega-os-public.git
claude  # everything just works

Getting framework updates

/update

Or manually: git config merge.ours.driver true && git fetch upstream && git merge upstream/master

Your personal data (active/, business/, products/, etc.) is protected by merge=ours rules in .gitattributes — framework updates never overwrite your files. The merge driver must be configured once locally (done automatically by /update and /setup). See GETTING_STARTED.md for full details.

ZIP users: Run engineering/scripts/check-updates.sh to check for new versions, or engineering/scripts/apply-update.sh to selectively update framework files.

Commands

Daily Operations:

Command	What it does
/goodmorning	Morning briefing with overnight cron results and suggested plan
/daily-scan	Morning digest — scans for stale, overdue, or unactioned items
/weekly-review	End-of-week retrospective with progress summary and priority updates
/news-briefing	AI-curated intelligence briefing for your domain
/dream	Generate a reflective prompt from the week's context
/connect	Find relevant connections in your network and draft introductions

Building & Creating:

Command	What it does
/project-kickoff	Scaffold a new product under products/
/write	Launch the content pipeline (Writer -> Editor -> Polisher)
/build-site	Build a website from concept to deployment (strategy → design → content → code → polish → QA)
/bug-triage	Diagnose and fix a reported bug
/deep-research	MECE-structured research (web, local codebase, or hybrid)
/generate-content	Generate short-form social content per channel schedule
/draw	Generate visual diagrams and images
/refract	AI use-case discovery: map a workflow, score opportunities, generate a custom demo

System Management:

Command	What it does
/setup	Guided onboarding wizard (re-run anytime to reconfigure)
/add-agent	Create a new custom agent
/add-skill	Create a new skill with all index updates
/improvement-audit	Deep system audit with rotating daily focus
/coherence	Invoke the Coherence perspective (harmonic awareness)
/workflow-review	Analyze workflow patterns and operational friction
/evolution-loop	Run the full self-improvement cycle (Evaluator → Coherence → Improver)
/framework-sync	Reconcile docs and sync after framework file changes (auto-triggered on commit)
/metrics-scan	Fetch PyPI, GitHub, and website metrics for your packages
/polish	Convert markdown to polished DOCX/PDF

Deployment & Updates:

Command	What it does
/update	Pull framework updates from the upstream public repo
/publish	Sync your framework changes to the public repo (maintainer only)

Directory Structure

mega-os/
  .claude/
    agents/            # 39 agent definitions in category subdirectories
      governance/      # overseer, governor, router, planner, pm, operator, sentinel, auditor, custodian
      knowledge/       # historian, librarian, summarizer, documenter, polisher, writer, editor
      technical/       # architect, engineer, executor, reviewer, qa, debugger, devops, security-expert, designer, visual-designer, api-designer
      business/        # strategist, marketer, seller, financier, proposal-writer, client-manager, content-strategist, growth-hacker
      evolution/       # improver, evaluator, coherence, parallax
      shared/          # system rules, collaboration protocol, templates, agent capability taxonomy
    skills/            # slash commands (29 total — setup, refract, build-site, deep-research, etc.)
  active/              # living task board (loaded every session)
    now.md             # current focus and active work
    priorities.md      # ordered priority list
    inbox.md           # incoming items awaiting triage
    blockers.md        # active blockers
    risks.md           # risk register
    improvements.md    # improvement proposals
  core/                # standards (17), templates (9), indexes, project history, trace data
  products/            # product codebases
  engineering/         # scripts, automations, infrastructure
  business/            # assets, clients, finance, marketing, operations

Agent Details

See AGENTS.md for the full list of agents, design philosophy, and key workflows.

The canonical registry of all agents lives at .claude/agents/REGISTRY.md.

How It Works

1. Session start: Claude reads your active/ files and picks up where you left off
2. During work: Agents handle specialized tasks (security reviews, content creation, business strategy, etc.)
3. Session end: State updates are committed to git — nothing is lost between sessions
4. Over time: The Evolution agents (Evaluator, Improver) propose system improvements based on usage patterns

Your OS learns and adapts. Decisions are recorded in core/history/. Improvements flow through a proposal/approval cycle. The system gets better as you use it.

License

MIT