# Lab Production Readiness Worksheet

Use this after completing a lab and before turning the pattern into production work.

Lab:
Reviewer:
Date:
Target product slice:

## 1. Lab Evidence

Baseline command:
Expected output:
Observed output:
Source files inspected:
Intentional failure path tested:

## 2. What The Lab Proved

[ ] The baseline command runs.
[ ] The success signal matches the lab.
[ ] One failure path is visible and controlled.
[ ] The main architecture boundary can be explained.

Notes:

## 3. What Production Would Add

[ ] Durable state or explicit stateless design.
[ ] Actor, tenant, resource, and tool authorization.
[ ] Typed tool schemas and side-effect labels.
[ ] Idempotency keys for externally visible writes.
[ ] Policy checks before tool calls, memory writes, and final answers.
[ ] Trace IDs, model/tool events, policy decisions, cost, latency, and stop reasons.
[ ] Eval gates for happy path, negative path, and trajectory behavior.
[ ] Deployment config, secret handling, rollback, and kill switch.
[ ] Human approval, escalation, cancellation, reviewer identity, and expiry.

Evidence:

## 4. Production Risk

What real data would this touch?
What real side effects could it cause?
What failure must fail closed?
What would require human approval?
What would block release?

## 5. Next Artifact

[ ] Framework selection ADR.
[ ] Tool capability review.
[ ] Policy enforcement checklist.
[ ] Observability and evals checklist.
[ ] Deployment walkthrough checklist.
[ ] Runbook.
[ ] Rollback plan.
[ ] Incident-derived eval fixture.

Owner:
Due date:
Blocking gaps:
Next action: