Skip to main content

Sprint 5+ Readiness Checklist

Gate between "we planned" and "we implement". Architect + PM Lead sign off jointly. Items marked ❌ block the kickoff.

Organizational​

  • Engineering owner assigned per module
  • Domain SMEs identified where relevant (finance SME for billing modules, security SME for auth, privacy SME for audience / user data)
  • On-call rotation set up
  • Stakeholders briefed (timeline, possible downtime, scope of change)

Documentation​

  • Module playbook (docs/architecture/modules/<module>-migration.md) reviewed and signed
  • Ontology gaps listed, queued for Stage 0
  • Relevant ADRs (0001–0010 range) are Accepted; owners know the rationale
  • Module runbook skeleton in place (copy of handoff template)

Infrastructure​

  • Monorepo scaffold exists (apps + packages as planned)
  • Core packages (ontology, db, contracts, harness) have hello-world in staging
  • apps/web, apps/api, apps/agent, apps/worker each deployable to staging
  • CI pipeline 10 steps all green on main: install / lint / typecheck / ontology validate / unit / integration / eval / build / traceability / bundle-size
  • Traceability workflow enabled on the implementation repo
  • Feature-flag system runnable: migration.<module>.mode togglable from dashboard

Platform capability​

  • SSO in staging — login works end-to-end
  • Postgres, Redis, queue service deployed to staging
  • LLM gateway reachable — can call at least two providers
  • Distributed tracing bridges web → api → db
  • Audit log writes succeed and are queryable
  • Dashboard platform configured — template can be instantiated per module

Migration tooling​

  • Bridge infrastructure (old ↔ new event forward) exists
  • Reconciliation script template (scripts/reconcile-<module>.ts) reviewed
  • reconciliation_diffs table created
  • Rollback procedure rehearsed at least once on a dummy module

Security & compliance​

  • Security review passed — PII tier tagging, authorization policies, secrets handling all sound
  • Privacy impact assessment (PIA) done for modules touching personal data
  • Regulatory / tax / legal review for modules where applicable (invoices, customer contracts, etc.)
  • Incident response playbook in place

Per-module Stage 0 gate (repeat per module)​

When a specific module is about to enter Stage 0, additionally:

  • Every item in module playbook § 9 Stage 0 checklist is actionable
  • Upstream module dependencies are at least in Stage 2 (or being migrated in parallel)
  • Change-freeze declared if the module is parameter-heavy or widely-read (e.g. products, user roles)
  • Module dashboard deployed; Block A and Block C panels have at least shadow-mode data
  • Module runbook sections 2.1–2.4 (four default incident playbooks) filled in

Sign-off​

RoleNameDate
Architect
PM Lead
Engineering Owner
On-call Lead
Security SME
Business / Domain SME

Anything still unchecked is a Sprint 5 blocker. "We'll fix it in Sprint 6" is how production incidents get planned.