enterprise-os

Enterprise OS

Enterprise-grade AI agent operating system for Claude Code. A 3-tier governance framework with 8 C-Suite agents, self-correction system, grade system, wave orchestration, plugin integration, and quality gates.

What is Enterprise OS?

Enterprise OS is a meta-system that gives any software project a complete AI-powered organizational structure:

• 8 C-Suite executives (CEO, CTO, CFO, CMO, CPO, COO, CLO, CHRO) with real-world leadership frameworks
• 9 team-level agents (team-leader, wave-supervisor, wave-watcher, system-grader, 4 specialist graders, actuary)
• 11 developer-level agents (feature-developer, qa, project-review, security-review, e2e-test-developer, plan-compliance, ux-tester, hotfix-developer, domain-resolver, conflict-analyzer, db-engineer)
• Self-correction system: 3-level agent self-improvement (self-score, lesson learning, evolution) across all 28 agents
• Plugin integration: Supports 28+ external plugins (trailofbits security suite, pg-aiguide, cartographer, humanizer, etc.)
• Quality gates: BUILD -> VALIDATE -> UX POLISH pipeline with parallel execution
• Grade system: 8-dimension evaluation (agents, hooks, skills, workflows, compatibility, liveness, intent, vocabulary)
• Wave orchestration: Parallel feature development across worktrees with real-time monitoring
• Migration safety: Pre-deploy checks, commit-gate and deploy-gate hooks preventing unsafe schema changes
• Enhanced security: FP Gate on security-review, security auto-scan hooks, secret-output-filter

Installation

claude plugin install enterprise-os

After installation, run the setup skill:

/setup

The setup skill will:

1. Scan your project structure
2. Guide you through domain vocabulary configuration
3. Generate a CLAUDE.md template with your project's conventions
4. Set up the docs/ directory structure for executive reports and task tracking
5. Configure hooks for your settings.local.json

Architecture

3-Tier Agent Hierarchy

L3 Executive (8 agents)
  CEO    — Binary Quality Gate, Specific Diagnosis, Cross-Functional Resolution
  CTO    — Architecture, /lead pipeline, /multi-lead parallel execution
  CFO    — Financial analysis, unit economics, SaaS metrics
  CMO    — Market strategy, growth loops, channel optimization
  CPO    — Product strategy, RICE prioritization, UX audit
  COO    — Operations, bottleneck analysis, mechanism design
  CLO    — Legal compliance, privacy, regulatory monitoring
  CHRO   — System health, agent quality, hook coverage

L2 Team (9 agents)
  team-leader          — Pipeline manager (BUILD -> VALIDATE -> UX POLISH)
  wave-supervisor      — Real-time wave monitoring, correction DMs
  wave-watcher         — Continuous observation during parallel wave execution
  system-grader        — Grade team orchestrator (8 dimensions)
  agent-quality-grader — Agent definition quality evaluation
  hooks-grader         — Hook enforcement coverage analysis
  skills-grader        — Skill ecosystem audit
  workflow-grader      — End-to-end workflow simulation
  actuary              — Quantitative analysis with statistical rigor

L1 Developer (11 agents)
  feature-developer    — Code implementation with engineering excellence
  qa                   — Quality gate (tests, build, lint, UI states)
  project-review       — Domain-specific code review
  security-review      — Security audit (RLS, auth, injection, XSS) with FP Gate
  e2e-test-developer   — Playwright E2E test authoring
  plan-compliance      — Scope drift detection
  ux-tester            — Live site UX testing via Playwright
  hotfix-developer     — Urgent fixes during wave execution
  domain-resolver      — Tier 3 merge conflict resolution
  conflict-analyzer    — Pre-merge conflict classification
  db-engineer          — Pre-deploy migration safety, RLS optimization, schema review

Development Pipeline

TRIVIAL:  BUILD -> QA + Security -> Done
STANDARD: BUILD -> VALIDATE -> E2E TEST (if new journey) -> UX POLISH (if UI) -> Done
MAJOR:    PLAN -> BUILD -> VALIDATE -> E2E TEST (if new journey) -> UX POLISH (if UI) -> PR

Self-Correction System (3 Levels)

All 28 agents are covered by the self-correction system:

Level	Name	Trigger	Action
A	Self-Score	Before submitting output	Agent scores own work 0-100; <70 triggers auto-rework (max 2x)
B	Lesson Learning	After validator review	Score delta >15 or CRITICAL finding generates a lesson; recent lessons auto-injected next run
C	Evolution	Same mistake 3x	Adds a check item to agent evolution file (CHRO reviews); max 20 items, 60-day TTL

Grade System (8 Dimensions)