CMOClaw → Splinter

V4 Architecture — 8 agents mapped to Splinter's /skills/:name/run endpoints. Compliance as middleware. Per-agent QMD memory.

Dual Runtime Mode

Same SKILL.md files power both systems

🧠 CMOClaw (OpenClaw)

  • 8 agents as sub-agents
  • QMD per agent namespace
  • memclawz for vector memory
  • WhatsApp dispatch
  • Compliance = pre-spawn check

🥷 Splinter (NestJS)

  • 8 agents as /skills/:name
  • Same SKILL.md files
  • Redis for session state
  • REST API dispatch
  • Compliance = interceptor

Agent → Splinter Endpoint Mapping

Click any row to see loaded skills

Splinter Controller

NestJS endpoints with compliance interceptor

// skills.controller.ts @Post('/skills/strategist/run') // 🎯 Research + briefs @Post('/skills/copy/run') // ✍️ Ad copy, headlines @Post('/skills/email/run') // 📧 Sequences, MJML @Post('/skills/creative/run') // 🎨 Visual assets @Post('/skills/builder/run') // 🛠️ Landing pages @Post('/skills/localisation/run') // 🌍 25 languages @Post('/skills/paid-media/run') // 📢 Channel deploy @Post('/skills/performance/run') // 📈 Analytics, A/B @Post('/skills/pipeline/run') // Full 6-wave campaign // Compliance = middleware interceptor (auto on all) @UseInterceptors(ComplianceInterceptor)

6-Wave Campaign Flow

Agent execution order with parallel waves and compliance middleware

Per-Agent QMD Memory

Each agent loads its own namespace from memclawz on spawn, updates on completion