Agentic Creator OS
frankxai/agentic-creator-os
SO para criadores alimentado por IA com 90+ skills auto-ativando
Ver no GitHubSistema operacional production-grade para assistentes de IA com 90+ skills, 65+ comandos e 38 agentes especializados que se auto-ativam conforme o contexto. Integra-se nativamente com Claude Code, Cursor, Windsurf e Gemini, oferecendo workflows reutilizáveis, hooks de segurança e roteamento inteligente de intenção.
Instalação
npx convez-ai frankxai/agentic-creator-os1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Automação de workflows de conteúdo (artigos, posts, SEO)
- Desenvolvimento full-stack com agentes especializados
- Roteamento inteligente de tarefas para múltiplos assistentes de IA
Veredito

Solução enterprise para quem precisa orquestrar múltiplos agentes de IA com segurança e contexto. Maduro tecnicamente, mas exige curva de aprendizado considerável e infraestrutura específica para máximo aproveitamento.
README
A tradução para o português deste README ainda está na fila.
Agentic Creator OS
The Operating System for AI-Powered Creators
One install. Any coding agent. 90+ skills, 65+ commands, 38 agents — auto-activating.

What Is the Agentic Creator OS?
A production-grade skill and agent system for AI coding assistants. It ships skills (domain knowledge), commands (reusable workflows), agents (specialized personas), and safety hooks — configured to auto-activate based on what you're working on.
Works with Claude Code, Grok Build (xAI CLI/TUI — full native harness + .grok/skills + .grok/hooks + subagents + MCP + grok-harness-adapter + Claude compat), Cursor, Windsurf, Gemini Code Assist, Antigravity, and any AI coding agent that reads markdown context files. Built on SIP (Starlight Intelligence Protocol).
You: "write a blog post about AI agents"
→ Agentic Creator OS detects: creation + blog
→ Auto-loads: content-strategy skill
→ Routes to: /article-creator workflow
→ Result: Guided article with SEO, images, social distribution
System Architecture

The Agentic Creator OS is structured in four layers, each auto-activating based on context:
flowchart LR
Intent["User intent"] --> Router["/acos smart router"]
Router --> Skills["Auto-activating skills"]
Router --> Commands["Slash workflows"]
Router --> Agents["Specialized agents"]
Skills --> Hooks["Safety hooks"]
Commands --> Hooks
Agents --> Hooks
Hooks --> Output["Audited creator output"]
Output --> Learning["Trajectory learning"]
Learning --> Router
| Layer | What It Does | Count |
|---|---|---|
| Skills | Domain knowledge modules that load automatically | 75+ |
| Commands | Reusable workflows triggered via slash commands | 35+ |
| Agents | Specialized personas with distinct expertise | 38 |
| Safety Hooks | Circuit breaker, audit trail, IAM, self-modify gate | 5 |
Architecture & Harness Role (Whole Stack)
ACOS (Agentic Creator OS) is the shared execution layer (skills + commands + agents + workflows + safety) used across the FrankX / Starlight ecosystem. It is harness-agnostic but provides first-class native adapters for the full fleet.
| Layer | Role | Key |
|---|---|---|
| SIP (Starlight Intelligence Protocol) | Cross-repo memory, MCP substrate (github/fs-starlight/git), sovereignty, attestation, vaults | starlightintelligence.org/protocol |
| ACOS | 90+ skills, 65+ commands, 38 agents, workflows, v10+ safety (IAM, circuit-breaker, audit, self-modify), auto-activation via skill-rules + excellence | This repo; junctions to ~/.claude + .grok compat |
| Grok Build (xAI) | TUI speed, subagents (explore/plan), MCP, image/video, .grok/skills + hooks native + full .claude compat. Personal excellence seeds (4 + 2 json) via grok-harness-adapter | adapters/grok/ + install --platform=grok; .grok-only grok-personal layer only |
| Claude Code | Canonical full surface (slash commands, hooks, IAM, skill-rules) | Direct .claude/ install |
| Other (Cursor, Windsurf, Gemini, AGY) | Context files + embedded skills/agents; delegation via multi-orchestrator | .cursorrules etc. |
| FrankX / Arcanea | Brand voice, personal creative, specific verticals (content, music, oracle, etc.) | Overlays + personal .grok/.claude partitions |
Core vs Grok-Personal vs Personal-Creative Partition (strict): See SHARING.md + SIP §5. The 4 excellence seeds live ONLY in .grok/ for Grok users ("a bit magical"). Everything else is core/shared via .claude/ junctions. Use sip-share-audit before cross moves. 5-fleet parity + God 99 gates (repo-mastery + gstack + santa/verification) on all work.
This separation + adapters make ACOS the universal substrate: one catalog, many harnesses, governed outputs.
Quick Start
Claude Code (Full Feature Set)
git clone https://github.com/frankxai/agentic-creator-os.git
cd agentic-creator-os
./install.sh
# Open any project with Claude Code
claude
/acos
Cursor / Windsurf
git clone https://github.com/frankxai/agentic-creator-os.git
cd agentic-creator-os
./install.sh --platform=cursor # or --platform=windsurf
This generates a .cursorrules or .windsurfrules file with all skills embedded.
Gemini Code Assist
git clone https://github.com/frankxai/agentic-creator-os.git
cd agentic-creator-os
./install.sh --platform=gemini
This generates a GEMINI.md context file for Gemini to read on session start.
Grok Build (xAI CLI/TUI) — Full Native Harness Support
git clone https://github.com/frankxai/agentic-creator-os.git
cd agentic-creator-os
./install.sh --platform=grok
- Generates
GROK.md(ACOS briefing + Frank DNA + excellence gates) + AGENTS.md pointer - Seeds
.grok/skills/(exact 4 grok-personal excellence seeds per SHARING.md + SIP §5: harness-integration, excellence-review, repo-mastery, multi-harness-orchestrator) + 2 excellence json hooks as.grok/hooks/ONLY +.grok/agents/(core ACOS/gstack/santa/verification via .claude/ junctions + compat) - Grok Build auto-loads project
.grok/(highest prio) +~/.grok/+ full~/.claude/compat (skills, commands, agents, hooks) - Native Grok features leveraged:
/skills, subagents (/taskwith personas + worktree + resume), MCP (github + fs-starlight + git), image/video gen, TUI speed - Run
grokin project →/hooks-trust→/skills grok-harness(or harness-integration / repo-mastery) or natural language ACOS tasks - Built-in excellence gates (God 99): repo-mastery (rules first), gstack (qa/browse/design-review), santa-method + verification-loop, plan-* reviews, cso
- Multi-harness: multi-harness-orchestrator auto-detects + emits exact delegation (e.g.
claude -p "..."with injected rules + DNA + partition filter)
This is the richest non-Claude experience — native skills/agents/hooks/subagents/MCP + full ACOS excellence substrate + 5-fleet parity. See adapters/grok/index.ts and .claude/skills/grok-harness/.
Any AI Coding Agent
git clone https://github.com/frankxai/agentic-creator-os.git
cd agentic-creator-os
./install.sh --platform=generic
Generates a CONTEXT.md file. Point your agent at it as a system prompt or project instructions file.
See QUICKSTART.md for detailed setup per platform
Core Concepts
The Agentic Creator OS is built around six foundational concepts. Understanding these unlocks the full system.
Skills — Auto-Activating Domain Knowledge
Skills are markdown files containing deep domain expertise. They activate automatically when the Agentic Creator OS detects relevant context in your work.
You start writing a blog post
→ skill-rules.json matches: "blog" + "write"
→ Auto-loads: content-strategy + seo-content-writer + schema-markup
→ Your AI agent now has SEO expertise, content structure, and metadata knowledge
| Skill Category | Examples | Trigger Patterns |
|---|---|---|
| Technical | TDD, React patterns, TypeScript, debugging | test, component, error |
| Creative | Brand voice, music mastery, book writing | blog, music, create |
| Infrastructure | Oracle Cloud, Kubernetes, Terraform | deploy, cloud, infra |
| Business | Product management, financial modeling | product, strategy, pricing |
Commands — Reusable Slash Workflows
Commands are multi-step workflows triggered by a single slash command. Each command orchestrates skills, agents, and tools in sequence.
| Command | What It Does | Steps |
|---|---|---|
/studio | Multimodal production | Brief → Route model → Craft → Generate (async) → Assemble |
/article-creator | Full blog pipeline | Research → Outline → Draft → SEO → Images → Social |
/factory | Mass content production | Batch input → Parallel agents → Quality gate → Export |
/spec | Feature specification | Requirements → Design → Task breakdown → Validation |
/infogenius | Visual intelligence | Topic → Research → 4K image generation → Metadata |
/create-music | Music production | Genre → Prompt engineering → Generation → Metadata |
/council | Strategic decisions | Multi-agent deliberation → Weighted synthesis → Verdict |
Agents — Specialized AI Personas
Each agent has distinct expertise, voice, and tool access. The Agentic Creator OS routes to the best agent based on task type.
| Agent | Specialty | When It Activates |
|---|---|---|
| Developmental Editor | Story structure, pacing, narrative architecture | Book/long-form writing |
| Music Producer | Suno prompts, genre production, audio engineering | Music creation |
| Technical Architect | System design, Oracle Cloud, enterprise patterns | Architecture tasks |
| UI/UX Expert | Design systems, accessibility, Figma integration | Frontend work |
| SEO Intelligence | Keywords, citations, structured data | Content optimization |
| Starlight Orchestrator | Multi-agent coordination, strategic decisions | Complex workflows |
Smart Router — Intent-Based Routing
The /acos command analyzes your request and routes to the optimal subsystem:
/acos "create an infographic about machine learning"
├── Intent detected: visual + educational
├── Agent selected: Visual Creator
├── Skills loaded: infogenius, design-excellence
├── Command chain: /research → /infogenius
└── Output: 4K infographic with research-grounded content
Multimodal Studio — Image · Video · Character
One connector, every modality. ACOS generates images, video, and consistent characters across 30+ frontier models through a single MCP — model-agnostic, brand-locked, and async by default.
/studio "hero image + 3 social cards + 5s teaser for the launch post"
├── Routes each shot to the right model (Soul/Flux → stills, Kling/Veo → motion)
├── Holds one character across the whole set via create_character → reuse ID
├── Submits assets in parallel, polls for results, downloads to canonical paths
└── Output: a coherent, brand-locked campaign — not disconnected one-off images
| Modality | Models (via Higgsfield MCP) | Differentiator |
|---|---|---|
| Image | Soul (4K), Flux, Seedream, Nano Banana | Brand tokens injected per asset |
| Video | Kling, Hailuo, Veo, Sora-class, DoP | Image→video first (cheaper, keeps composition) |
| Character | Soul character training | One reference, identical across an entire series |
Skill: multimodal-studio · Agent: Multimodal Director · Commands: /studio, /generate-video. Connect: claude mcp add --transport http --scope user higgsfield https://mcp.higgsfield.ai/mcp. Stays vendor-agnostic — any MCP filling ~~image generation / ~~video generation works (CONNECTORS.md). Full guide + the ACOS-vs-Antigravity positioning: docs/multimodal-studio.md.
Safety Hooks — Autonomous Protection
Five systems that run automatically to keep the Agentic Creator OS reliable:
| Hook | What It Prevents |
|---|---|
| Circuit Breaker | Stops the agent from brute-forcing broken approaches (3→warn, 5→restrict, 8→block) |
| Audit Trail | Append-only JSONL log of every operation — tamper-proof ground truth |
| Agent IAM | Role-based scoping — a content writer can't modify build configs |
| Self-Modify Gate | Snapshots state before config changes, auto-reverts if intelligence drops |
| Quality Gate | Blocks edits to secrets files, warns on production file modifications |
Self-Learning — Gets Smarter Every Session
The Agentic Creator OS tracks tool sequences and outcomes across sessions, then injects successful patterns as context for similar future tasks. No external API — pure local intelligence.
v10 — Autonomous Intelligence
v10 introduces 5 safety-first autonomous systems that make your AI coding agent smarter over time:
1. Experience Replay
The agent remembers what worked. Past successful tool sequences are injected as context for similar new tasks.
2. Agent IAM (Identity & Access Management)
Per-profile scoping — a content writer can't modify your build config, a music producer can't run shell commands. 6 profiles with tool and directory restrictions.
3. Confidence Circuit Breaker
Tracks failures per file. After 3 failures: warn. After 5: restrict. After 8: block. Prevents the agent from brute-forcing broken approaches.
4. Conservative Self-Modify Gate
Before changing its own config, ACOS snapshots the current state and measures intelligence score. If the score drops more than 5 points after a change: auto-revert.
5. Immutable Audit Trail
Append-only JSONL logging of every tool use, IAM decision, gate verdict, and config change. The ground truth for what happened in a session.
/acos (Smart Router)
|
+-------------+-------------+
| | |
Commands (35+) Skills (75+) Agents (38)
| | |
+------+------+------+------+
| |
+------+------+ +--+--+
| | | |
Hooks (v10) Workflows MCP Servers
|
+---------+---------+---------+
| | | |
Circuit Audit Self-Mod Agent
Breaker Trail Gate IAM
What's Included
Commands (35+)
| Category | Count | Examples |
|---|---|---|
| Creation | 12 | /studio, /generate-video, /article-creator, /create-music, /infogenius, /factory |
| Strategy | 6 | /starlight-architect, /council, /research, /plan-week |
| Development | 4 | /spec, /nextjs-deploy, /ux-design, /automation-dev |
| System | 5 | /acos, /inventory-status, /mcp-status, /publish |
| Quality | 3 | /review-content, /classify-content, /polish-content |
Commands are Claude Code slash commands. On other platforms, describe the task and ACOS skills guide the agent to the same workflows.
Skills (75+ Auto-Activating)
Skills are domain knowledge modules that activate automatically based on what you're doing. No manual invocation needed.
| Category | Count | Examples |
|---|---|---|
| Technical | 25+ | TDD, systematic-debugging, mcp-architecture, react-nextjs-patterns |
| Creative | 10+ | multimodal-studio, frankx-brand, suno-ai-mastery, excellence-book-writing |
| Content | 15+ | content-strategy, social-media-strategy, video-production |
| Oracle/Cloud | 5+ | oci-services-expert, oracle-database-expert, oracle-ai-architect |
| Business | 5+ | product-management, coaching-program, financial-modeling |
| System | 15+ | swarm-orchestration, agentic-jujutsu, planning-with-files |
| Personal | 5+ | spartan-warrior, gym-training, health-nutrition |
How Auto-Activation Works:
skill-rules.json → 22 pattern rules
"tests" + "component" → loads test-driven-development + react-patterns
"blog" + "write" → loads content-strategy + seo-content-writer
"deploy" + "vercel" → loads vercel-deployment + nextjs-best-practices
Agents (38 Specialized)
| Domain | Agents | Examples |
|---|---|---|
| Writing | 8 | Developmental Editor, Line Editor, Content Polisher, Story Architect |
| Strategy | 5 | Starlight Orchestrator, Visionary, Oracle AI Architect |
| Design | 4 | UI/UX Expert, Website Builder, Frankx Content Creation |
| Production | 5 | Multimodal Director, Music Production, Sonic Engineer, Nano Banana Image Gen |
| Business | 4 | Product Launch, Coaching Program, Product Development |
| Technical | 5 | Technical Architect, Oracle Database, OpenAI AgentKit |
| Publishing | 6 | Master Story Architect, Character Psychologist, Sensitivity Reader |
| System | 3 | Skill Expert, Project Discovery, Rapid Content |
v10 Safety Hooks
| Hook | Trigger | Purpose |
|---|---|---|
| Circuit Breaker | PostToolUse (failure) | Track failures per file, escalate at thresholds |
| Audit Trail | All tool use | Append-only JSONL logging |
| Self-Modify Gate | Config changes | Snapshot + validate intelligence score |
| Agent IAM | PreToolUse | Enforce per-profile tool/path scopes |
| Quality Gate | PreToolUse (Edit/Write) | Block edits to secrets, warn on production files |
How It Works
The /acos Smart Router

Everything starts with /acos. It parses intent and routes to the best subsystem:
/acos "write a blog post about AI agents"
→ Detects: creation + blog
→ Routes to: /article-creator
→ Auto-loads: content-strategy skill
→ Result: Guided article creation workflow
For multi-domain requests, it chains commands:
/acos "Create a blog post about AI music with images and social posts"
→ /research "AI music production"
→ /article-creator (with research context)
→ /infogenius (hero image)
→ /generate-social (distribution)
Orchestration Patterns
| Pattern | Description | Best For |
|---|---|---|
| Pipeline | Sequential: Research → Plan → Create → Optimize → Publish | Blog posts, newsletters |
| Parallel | Concurrent agents with synthesis | Code review, multi-analysis |
| Weighted Synthesis | Expert voting with percentage weights | Strategic decisions |
Self-Learning

The Agentic Creator OS learns from every session through a four-stage cycle:
| Stage | What Happens |
|---|---|
| 1. Session Start | Creates a trajectory record for the current session |
| 2. Track Operations | Monitors tool sequences, file changes, and outcomes |
| 3. Extract Patterns | Identifies successful workflows and tool combinations |
| 4. Apply Intelligence | Injects learned patterns as context for future sessions |
Session Start → Create trajectory record
Session Active → Track operations and outcomes
Session End → Extract patterns → Store learnings
Future Sessions → Apply accumulated intelligence
Multi-Platform Install
The install script detects your platform and configures accordingly:
./install.sh # Auto-detect platforms
./install.sh --platform=claude # Claude Code only
./install.sh --platform=cursor # Cursor only
./install.sh --platform=windsurf # Windsurf only
./install.sh --platform=gemini # Gemini Code Assist only
./install.sh --platform=grok # Grok full (GROK.md + .grok/ seeds from grok-harness-adapter)
./install.sh --platform=generic # Any agent (CONTEXT.md)
./install.sh --platform=all # All detected platforms
What Gets Installed Per Platform
| Component | Claude Code | Grok Build (xAI) | Cursor | Windsurf | Gemini | Generic |
|---|---|---|---|---|---|---|
| Skills (knowledge) | ~/.claude/skills/ | .grok/skills/ + GROK.md + ~/.claude compat (via grok-harness-adapter seeds) | .cursorrules | .windsurfrules | GEMINI.md | CONTEXT.md |
| Commands (workflows) | ~/.claude/commands/ | Context + native /skills + subagents | — | — | — | — |
| Agents (personas) | ~/.claude/agents/ | .grok/agents/ + subagents | .cursorrules | .windsurfrules | GEMINI.md | CONTEXT.md |
| Hooks (safety) | settings.json + acos/hooks | .grok/hooks/ (trust with /hooks-trust) + excellence gates | — | — | — | — |
| Auto-activation | skill-rules.json | harness-integration + excellence-review seeds (gstack/santa/verification) | — | — | — | — |
| MCP servers | Optional | — | — | — | — |
Claude Code gets the full feature set including slash commands, lifecycle hooks, and auto-activation rules. Other platforms get skills and agent definitions embedded in their context files.
Directory Structure
agentic-creator-os/
├── .claude/
│ ├── commands/ # 35+ slash commands
│ ├── skills/ # 75+ auto-activating skills
│ ├── agents/ # 38 specialized agents
│ ├── hooks/ # v10 safety hooks
│ │ ├── circuit-breaker.sh
│ │ ├── audit-trail.sh
│ │ ├── self-modify-gate.sh
│ │ └── quality-gate.sh
│ ├── agent-iam.json # Role-based access control
│ ├── skill-rules.json # 22 auto-activation rules
│ └── hooks.json # Hook lifecycle config
│
├── adapters/ # Platform adapters (TS context + seed generators)
│ ├── grok/ # grok-harness-adapter: index.ts (isGrok, parse, generateGrokContext, getGrokSeeds, installGrokPlatform), GROK excellence seeds
│ ├── opencode/ # OpenCode integration adapter
│ ├── cursor/ # .cursorrules generator (planned)
│ ├── windsurf/ # .windsurfrules generator (planned)
│ ├── gemini/ # GEMINI.md generator (planned)
│ └── generic/ # CONTEXT.md generator (planned)
│
├── departments/ # Agent team definitions
│ ├── content/
│ ├── design/
│ ├── dev/
│ ├── marketing/
│ └── business/
│
├── mcp-servers/ # MCP integrations (optional)
├── templates/ # Reusable templates
├── workflows/ # Workflow definitions
├── docs/infographics/ # Visual documentation
├── install.sh # Multi-platform installer
├── CLAUDE.md # Claude Code context
└── package.json # v10.1.0
The Starlight Connection
ACOS is the Claude Code implementation of the Starlight Intelligence System — a universal, platform-agnostic framework for multi-agent orchestration.
Starlight Intelligence System (Framework)
├── 5-layer cognitive architecture
├── 7 council agents with emergent leadership
├── 6 orchestration patterns
└── Platform adapters
└── ACOS (Claude Code + Grok full harness via grok-harness-adapter + Multi-Platform)
├── 35+ commands routed through /acos
└── Grok: adapters/grok/index.ts + install.sh --platform=grok (GROK.md + .grok/ seeds: excellence, repo-mastery, gstack gates)
├── 38 agents aligned to Starlight council
├── 75+ auto-activating skills
├── v10 safety systems (IAM, circuit breaker, audit)
└── Self-learning via trajectory patterns
Design Principles
- Configuration over Code — Markdown + JSON + AI's native abilities. Zero install friction.
- Commands over Prompts — Reusable workflows beat one-off prompts. Build once, use forever.
- Agents over Chat — Specialized expertise beats general conversation.
- Files over Memory — Persistent artifacts beat ephemeral threads. Everything is git-trackable.
- Progressive Disclosure — Load only what you need. ~100 tokens metadata before 5K skill definition.
- Open and Forkable — MIT licensed. Every component replaceable.
Plugin Marketplace
Extend ACOS with installable domain plugins from agentic-creator-skills:
| Plugin | Purpose |
|---|---|
| core | Session management, workspace memory, task tracking |
| content-engine | Content creation, voice enforcement, publishing |
| visual-studio | Style-governed image generation, 7-gate quality filter |
| music-lab | AI music production, Suno prompt engineering |
| brand-architect | Brand voice system, tone adaptation, terminology |
| product-launcher | Product specs, sprint management, deployment |
| intelligence | Intelligence scoring, pattern analysis, reporting |
| design-excellence | Design system governance, 5-gate QA, Figma integration |
# Install plugins (Claude Code)
claude plugin marketplace add frankxai/agentic-creator-skills
claude plugin install core content-engine design-excellence
ACOS = the operating system. Plugins = the app store. Install only what you need.
Version History
| Version | Date | Highlights |
|---|---|---|
| v11.0 | Feb 2026 | Plugin ecosystem, design swarm, v11 skills absorbed from acos-intelligence-system |
| v10.1 | Feb 2026 | Deep clean — removed claude-flow artifacts, honest metrics, clean repo |
| v10.0 | Feb 2026 | Autonomous Intelligence — 5 safety systems, multi-platform, Agent IAM |
| v8.0 | Feb 2026 | Starlight integration, GitHub release assets, swarm orchestration |
| v7.0 | Feb 2026 | Visionary agent, premium infographics, expanded commands |
| v6.0 | Jan 2026 | Smart router (/acos), auto-activation, hooks |
| v5.0 | Jan 2026 | 7-pillar architecture, 62 skills, swarm topologies |
| v4.0 | Jan 2026 | Initial release, 14 commands, basic skills |
Credits & Attribution
ACOS builds on work from the open-source community:
| Project | Author | What We Learned |
|---|---|---|
| claude-flow | @ruvnet | Swarm orchestration, hierarchical topologies |
| wshobson/agents | @wshobson | Plugin architecture, modular skills |
| obra/superpowers | @obra | Progressive disclosure, token-efficient loading |
| diet103/claude-code-infrastructure-showcase | @diet103 | skill-rules.json auto-activation pattern |
| ChrisWiles/claude-code-showcase | @ChrisWiles | Hook automation patterns |
| Pimzino/claude-code-spec-workflow | @Pimzino | Spec-driven development |
| github/github-mcp-server | GitHub | Official GitHub MCP server |
| agentic-jujutsu | @ruvnet | Self-learning trajectory patterns |
Built on: Claude Code by Anthropic, Model Context Protocol
Related Projects
- agentic-creator-skills — Plugin marketplace (8 domain plugins)
- Starlight Intelligence System — The universal framework powering ACOS
- Arcanea — AI-native creative platform built with ACOS
License
MIT — Use it, fork it, build your own OS with it.
Agentic Creator OS v11.0 — The Operating System for AI-Powered Creators
90+ Skills | 65+ Commands | 38 Agents | 8 Plugins | Multi-Platform | Self-Learning
Built by FrankX — AI Architect & Creator