ADR-0005proposed

AIAA OS — TurboClaw integration for per-student OpenClaw deployment

Context and Problem Statement

AIAA teaches agency owners to use AI agents, but students struggle with deployment infrastructure. Three existing repos contain pieces of the solution: TurboClaw (per-user deployment), Agentcy-OS (gateway engine with 29 channel extensions), and ClawsOS (145 agency directives + dashboard). These need to be unified into a single product that gives each AIAA student their own AI agent with pre-loaded agency workflows.

Decision

Fork turbostarter/openclaw (TurboClaw) as the deployment foundation and integrate it with AIAA-specific content:

  1. Custom Docker image — Build on official OpenClaw image, bake in 145 agency directives from ClawsOS + 286 skill bibles from Agentcy-OS + AIAA-branded AGENTS.md/SOUL.md templates
  2. Extended config generation — Expand getGatewayConfig() to inject agency context files (profile, brand voice, services, ideal client) alongside the standard openclaw.json
  3. Agency setup wizard — Dashboard form that captures agency details and writes context files to the instance STATE_DIR before container start
  4. Workflow dashboard — Port ClawsOS workflow browser (categories, detail pages, “Run” button) into TurboClaw’s Next.js app

Architecture: TurboClaw control plane → Docker container per student → OpenClaw with pre-loaded AIAA content → Connected via Telegram/Discord/Slack

Consequences

  • Good, because students go from signup to running agent in under 60 seconds
  • Good, because 145 pre-built workflows eliminate the “what do I do with this” problem
  • Good, because TurboClaw already handles the hard infra (SSH, Docker, Caddy, HTTPS)
  • Bad, because maintaining a custom Docker image adds build/release overhead
  • Bad, because shared API keys across instances need rate limiting and cost tracking
  • Neutral, because the product can be monetized as a standalone SaaS or bundled with AIAA enrollment

Implementation Plan

Phase 1: POC (1-2 days)

  • Fork turbostarter/openclaw
  • Build custom Docker image with directives
  • Deploy one test instance on VPS
  • Verify: chat works, agent has agency context

Phase 2: Agency Setup (2-3 days)

  • Setup wizard in dashboard
  • Context file generation and injection

Phase 3: Workflow Dashboard (3-5 days)

  • Port ClawsOS components to Next.js
  • Chat panel integration

Phase 4: Billing (3-5 days)

  • Stripe subscriptions, usage metering

Phase 5: Launch (2-3 days)

  • Landing page, templates, monitoring

Affected repos

  • turbostarter/openclaw (fork) → core deployment
  • lucassynnott/clawsos → directives source
  • lucassynnott/Claws-Agentcy-OS → skill bibles source

Verification

  • Custom Docker image builds and starts
  • Instance deploys from dashboard with agency context
  • Agent responds with agency-aware context
  • 145 workflows browsable in dashboard
  • Stripe checkout → auto-deploy flow works end-to-end

More Information

Full plan at /nas/vault/people/workspaces/johnny/plans/aiaa-os-plan.md