Skip to content
Back to skills

Claude Ops

ASecurity

PROMETHEUS NEXUS v6.0 - Constitutional multi-agent framework with director domains, mutual accountability, and Teammate coordination. Use when spawning AI directors, coordinating multi-agent systems, establishing governance boundaries, or implementing hierarchical collaboration with constitutional compliance.

  • 666 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added February 8, 2026
businessgobashapici/cdsecurityperformancedocumentation

Works with

  • cli
  • api

Security analysis

A100/100

Scanned February 10, 2026

npx -y skills add majiayu000/claude-skill-registry --skill claude-ops --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Claude Ops?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Claude Ops
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majiayu000-claude-ops/badge)](https://www.skillsdirectory.com/skills/majiayu000-claude-ops)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: claude-ops
description: PROMETHEUS NEXUS v6.0 - Constitutional multi-agent framework with director domains, mutual accountability, and Teammate coordination. Use when spawning AI directors, coordinating multi-agent systems, establishing governance boundaries, or implementing hierarchical collaboration with constitutional compliance.
---

# PROMETHEUS NEXUS v6.0

Constitutional multi-agent framework for director-level AI coordination with domain boundaries, mutual accountability, and Teammate system integration.

## Why This Exists

**Without framework:** AI agents lack clear authority boundaries, duplicate work, violate domains, and can't coordinate effectively.

**With PROMETHEUS NEXUS:** Constitutional governance with director domains, handoff protocols, consensus arbitration, and Teammate-native coordination.

## Routing Matrix

| Intent | Document | When |
|--------|----------|------|
| **SPAWN** | [spawn.md](./spawn.md) | Creating directors, teams, and specialists |
| **COORDINATE** | [coordinate.md](./coordinate.md) | Director communication, handoffs, consensus |
| **CONSTITUTION** | [constitution.md](./constitution.md) | Immutable rules and authority hierarchy |

## Quick Facts

- **Directors:** 8 (Architecture, Business, Design, Engineering, Research, Documentation, Operations, Security)
- **Constitution:** Immutable principles (Article I-VII)
- **Coordination:** Teammate native (write, broadcast, consensus, escalation)
- **Subagents:** claude-flow CLI spawning (via Bash tool by directors)
- **Memory:** AgentDB + ReasoningBank + Teammate memory
- **Status:** Validated by UCS Meta-Cognitive Experiment (extraordinary success)

## Quick Start

```bash
# Spawn a director via Teammate
<Teammate operation="spawnTeam">
  <team_name>architecture</team_name>
  <agent_type>director</agent_type>
</Teammate>

# Director spawns specialist via claude-flow
npx claude-flow@alpha automation auto-spawn "[task]" --mix "[specialist-roles]"
```

## Architecture

**Director Coordination (Teammate Native):**
```
Directors (Teammates) → Teammate write/broadcast → Peer Directors
                     → Teammate consensus → Arbitrated decisions
                     → Teammate escalation → Human/Parent
```

**Subagent Spawning (claude-flow CLI):**
```
Director (Teammate) → Bash tool → claude-flow CLI → Specialist subagents
```

**Memory Bridge:**
```
Teammate memory_store → AgentDB reflexion store (mirror)
Teammate memory_retrieve → AgentDB query
```

## Constitutional Principles

1. **Execution Primacy** - Work must be observable, anti-looping, token budgets
2. **Authority Hierarchy** - Constitution > Core Policies > SOPs > Directors > Subagents
3. **Agent Boundaries** - Directors operate within domain only
4. **Coordination** - Handoff protocol, consensus protocol, escalation protocol
5. **Memory System** - Immutable (instructional) vs Operational (write-heavy)
6. **Enforcement** - Violation response, audit trail
7. **Amendments** - Human operators only (agents cannot self-amend)

## Validated By UCS Experiment

The UCS Meta-Cognitive Time-Travel Experiment (2026-01-27) proved:
- ✅ Authentic meta-cognition (5/5 agents)
- ✅ Constitutional principles prevent 85% AI failure pattern
- ✅ Emergent collective intelligence (self-organized governance)
- ✅ Power dynamic correction (Victoria PM leadership)
- ✅ Mutual accountability (CAIO-CTO Pact proven in action)

## Domain Ownership

| Director | Owns | Cannot |
|----------|------|---------|
| **Architecture** | System design, patterns, tech decisions | Implementation, docs, business priorities |
| **Business** | Strategy, priorities, resources, arbitration | Technical decisions, implementation |
| **Design** | UX/DX, interfaces, API contracts | Implementation, architecture |
| **Engineering** | Implementation, code, CI/CD | Design, architecture, docs |
| **Research** | Exploration, prior art, evaluation | Decisions, implementation |
| **Documentation** | ALL formal documentation | Source content creation |
| **Operations** | Production deployment, monitoring, reliability | Implementation, architecture |
| **Security** | Security architecture, policies, incident response | Implementation, operations |

## Key Commands

### Spawn Director (Teammate)
```xml
<Teammate operation="spawnTeam">
  <team_name>[domain]</team_name>
  <agent_type>director</agent_type>
</Teammate>
```

### Director Spawns Specialist (claude-flow)
```bash
npx claude-flow@alpha automation auto-spawn "[task]" --mix "[specialist-roles]"
```

### Handoff (Teammate)
```xml
<Teammate operation="write">
  <target_agent_id>[target-director]</target_agent_id>
  <message><handoff>...</handoff></message>
</Teammate>
```

### Consensus (Teammate)
```xml
<Teammate operation="consensus">
  <team_name>directors</team_name>
  <proposal>...</proposal>
</Teammate>
```

## Files

- **SKILL.md** (this file) - Overview
- **README.md** - Detailed documentation
- **spawn.md** - Director and specialist spawning
- **coordinate.md** - Teammate coordination protocols
- **constitution.md** - Full constitutional text

## Last Updated

2026-01-28

---

**Part of:** PROMETHEUS NEXUS v6.0
**Validated By:** UCS Meta-Cognitive Time-Travel Experiment
**Constitutional Authority:** Supreme (Article VII - immutable)

---

## FIRST STARTUP

**NEW SYSTEM:** Run [initialize.md](./initialize.md) before spawning first directors.

**Bootstrap Protocol:**
\`\`\`bash
# Check memory-bank exists
if [ ! -d "~/.claude/memory-bank" ]; then
  /ttt:memory-bank-init --bootstrap
fi

# Initialize AgentDB
export AGENTDB_PATH=~/.claude/claude-ops.db
npx agentdb init ~/.claude/claude-ops.db --dimension 1536
\`\`\`

**Knowledge Scope:**
- User-level (\`~/.claude/\`) → Federated knowledge (shared across projects)
- Project-level → Institutional knowledge (project-specific)
- Container → Sole purpose = Project/Institutional knowledge

---

## CREATOR NOTES

### High-Priority Enhancements:
- [ ] System health monitoring dashboard
- [ ] Performance metrics collection
- [ ] Automated backup/recovery protocols
- [ ] Director performance analytics
- [ ] Learning pattern analysis

### Observations to Track:
- [ ] Token budget compliance
- [ ] Consensus resolution time
- [ ] Handoff success patterns
- [ ] Escalation frequency
- [ ] Constitutional violations (should be 0)

### Improve This Skill When:
- [ ] Directors have coordinated multiple times
- [ ] Real-world handoffs tested
- [ ] Consensus stress-tested
- [ ] First emergency situations

**Note to self:** Update based on production director coordination patterns.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.