Courier TMS / Launch Gate 01Fixed scope · engineering review
For Next.js + React Native + Supabase operations

Prove the fleet can keep moving.

A 48-hour production-readiness audit for a courier TMS already carrying live operational risk. The output is evidence: what blocks a safe rollout, what can wait, and exactly how to run a 2–3 driver canary.

Confirm the £200 scopeSee the launch gate
Three signals, kept separate

“The API is up” is not the same as “dispatch is working.”

The audit separates infrastructure health from application errors and the actual freshness of the courier workflow. That prevents a green dashboard from hiding a stalled route.

SIGNAL A

Availability

Can drivers and dispatch reach the service?

SIGNAL B

Exceptions

Which code path is failing, and for whom?

SIGNAL C

Freshness

Did each active driver's latest GPS update arrive inside the agreed window?

The exact 48-hour route

From access boundary to go/no-go.

HOUR 00–16

Auth boundary

Map roles, RLS policies and ownership checks using redacted schema and a client-provided test account.

HOUR 16–34

Recovery & flow

Review migrations, rollback, backup evidence, GPS-write paths and workflow-freshness monitoring.

HOUR 34–44

Retention proof

Trace the 90-day job, its row counts, rejected records and the follow-up query that proves the boundary.

HOUR 44–48

Launch pack

Deliver the P0/P1/P2 register, canary plan, rollback triggers and incident runbook.

What arrives

A handoff an operator can use.

Evidence-backed report

  • P0: blocks a safe canary
  • P1: fix before broad fleet trust
  • P2: lower-urgency hardening
  • File, screenshot, redacted log or client-run query for every finding

Canary & runbook

  • 2–3 volunteer drivers for 3–5 business days
  • Explicit success and rollback criteria
  • Availability, exception and GPS-freshness response steps
  • Retention-job failure procedure

Safe access boundary

No passwords. No service-role secrets. No unrestricted production access. No real customer or driver PII export.

Use a read-only repository role, redacted evidence, staging and a non-production test account.

Claim boundary

This is an engineering production-readiness review. It is not penetration testing, legal advice, GDPR certification or a security certification.

Implementation is priced only after the evidence-backed backlog is reviewed.

Fund the audit. See the risks. Then decide on maintenance.

Send the stack summary, current monitoring, staging status and the five flows you cannot afford to lose. The 48-hour window starts only after scope, evidence boundaries and payment are confirmed in writing.

Request the fixed-scope intake