⚡ Atlas Command Center

PAI + Atlas Agent & Capability Architecture

Daily Budget: $10.00
Active Agents: 4 Atlas + 8 PAI
Skills Available: 18

Atlas Agents

Elvis

Atlas

Code engineering & system architecture

Model: Claude Opus 4.6
Budget: Unlimited (subscription)
Invocation: claude (local session)
View Capabilities
  • • Full-stack engineering (Python, TypeScript, Bash, SQL)
  • • PAI Algorithm integration (7-phase execution)
  • • Git operations, PR creation, deployment
  • • MCP: Telegram, Gmail, Google Calendar
  • • Session persistence management

LORE

Atlas

Research & competitive intelligence

Model: DeepSeek V3.2
Budget: $10/day (shared)
Invocation: OpenClaw → agent: lore
View Capabilities
  • • Multi-mode research (Quick/Standard/Extensive/Deep)
  • • Parallel agent research (up to 5 agents)
  • • 242+ Fabric patterns for content enhancement
  • • WebSearch, WebFetch, PDF extraction
  • • Knowledge base curation

RENDER

Atlas

Document extraction & pipeline blueprints

Model: DeepSeek V3.2
Budget: $10/day (shared)
Invocation: OpenClaw → agent: render
View Capabilities
  • • Reverse-engineer artifacts into blueprints
  • • Extract structured data from PDFs/reports
  • • Convert manual workflows to automation specs
  • • Confidence scoring and validation
  • • LRB Circle integration (LORE → RENDER → Bob)

Flux

Atlas

n8n workflow automation

Model: DeepSeek V3.2
Budget: $10/day (shared)
Invocation: OpenClaw → agent: flux
View Capabilities
  • • Design production-ready n8n workflows
  • • Generate workflow JSON from natural language
  • • Debug existing workflows
  • • Validate logic and error handling
  • • n8n REST API bridge integration

PAI Agent Types

Task Tool Subagent Types

Internal workflow agents (no unique voice)

Development Agents

Architect — System design & architecture
Designer — UX/UI design
Engineer — Code implementation

Research Agents

ClaudeResearcher — Claude-based research
GeminiResearcher — Gemini-based research
GrokResearcher — Grok-based research

Specialized Agents

Explore — Codebase exploration
Plan — Implementation planning
QATester — Quality assurance
Pentester — Security testing
general-purpose — Custom agents via ComposeAgent

Named Agents

Persistent identities with ElevenLabs voices

Serena Blackwood — Architect | Premium UK Female
Marcus Webb — Engineer | Premium Male
Rook Blackburn — Pentester | Enhanced UK Male
Ava Sterling — Claude Researcher | Premium US Female
Alex Rivera — Gemini Researcher | Multi-perspective

Custom Agents

Dynamic composition via ComposeAgent

Composed on-the-fly from trait combinations:

Expertise security, legal, finance, technical, research, creative...
Personality skeptical, enthusiastic, analytical, bold, empathetic...
Approach thorough, rapid, systematic, exploratory, adversarial...
Each unique trait combination maps to a different ElevenLabs voice

Agent Team Patterns

When to Use Council (Multi-Agent Debate)

  • • Need multiple perspectives on a decision
  • • Evaluating trade-offs
  • • 3-5 agents with different viewpoints
  • • Time: 2-5 minutes

When to Use Deep Investigation

  • • Comprehensive landscape mapping
  • • Progressive entity deep-dives
  • • 12+ agents in parallel
  • • Time: 30-60 minutes

When to Use Custom Agents

  • • Need specific expertise combinations
  • • Want adversarial viewpoints
  • • Compose via trait combinations
  • • Each gets unique voice + personality

Research Delegation Pattern

  • • Elvis needs data → delegates to LORE
  • • LORE runs Standard (3 agents) or Extensive (12 agents)
  • • Report returns to Elvis for decision
  • • Time: 5-30 minutes depending on mode