Skip to main content

​
Goal

Enable OpenClaw-driven execution without compromising safety, reliability, or documentation quality.

​
Locked Decisions (2026-02-20, confirmed 2026-02-21)

  1. Channels:
    • WhatsApp for external influencer/client communication
    • Telegram for internal manager approvals and operational updates
  2. Runtime ownership:
    • Render is primary app runtime
    • Supabase is primary Postgres/Auth runtime
    • Hostinger OpenClaw node is primary control plane
  3. Node topology:
    • dedicated Hostinger VPS node is primary
    • MacBook node is fallback only
  4. Vercel scope:
    • no active Vercel deployment path is maintained
    • not canonical production infrastructure
  5. Exec policy:
    • exec/process is denied by default
    • explicit manager confirmation is required for one-off execution
  6. Initial write scope:
    • docs and non-production code paths
  7. External communication policy:
    • use neutral role language; do not expose internal relationship context

​
Role Model

  • Manager agent: planning, routing, merge gating, risk ownership
  • Support agent: implementation, testing, docs updates, handoff

​
Phased Rollout

  1. Foundation
    • standardize runbooks and docs contracts
    • enforce required checks (pnpm check, tests, docs)
  2. Controlled execution
    • manager-approved task routing only
    • evidence-based handoff format required
  3. AI feature expansion
    • template enhancement and analysis workflows through explicit API contracts
    • audit trail for agent-originated writes
  4. Continuous governance
    • weekly health review for code, CI, docs, security posture

​
Quality and Stability Program

  1. Required gates on all AI-authored changes:
    • pnpm check
    • pnpm test:unit
    • pnpm test:integration
    • pnpm docs:check
    • pnpm openclaw:check
  2. Active quality backlog:
    • reduce any in high-risk backend routes
    • eliminate React act(...) warning noise in tests
    • reduce non-actionable Node warning noise in test logs

​
Non-Negotiable Gates

  • no merge without docs updates for behavior changes
  • no merge on auth/public/share changes without tests
  • residual risk and rollback note required

​
Target Outcomes

  • faster safe delivery
  • fewer regressions in high-risk paths
  • consistent onboarding for new human and AI contributors

​
Source Docs