# Deliver as an engineering pod — free clinic brief

AI-native training by LockedIn Labs · Edition 1, 28 October 2026

Use the [Verified event handoff prepared session](https://training.lockedinlabs.ai/training/guides?starter=verified-event-handoff#guide-session-title). Preview its exact current labs before applying it to your agenda. The plan contains two 30-minute practice and 15-minute receiving phases: 90 entered minutes, with setup, reading and breaks separate. Prepare Python 3.11+ and both standard-library runners in disposable folders. No package installation, model or remote service is needed for these two exercises.

## Run two receiving turns

| Planned minutes | Roles and task | Evidence |
| --- | --- | --- |
| 0–30 | A builds; B receives. Predict and run reserialized body bytes, a valid delivery aged 301 seconds and a fresh repeat. | Exact commands, actual MAC/time/parse/transaction results and final receipt/effect state. |
| 30–45 | B explains the verified-byte boundary and attempts the changed case without author takeover. | B's own action, missing evidence and assistance; source/application owner's next test. |
| 45–75 | Swap roles. B builds; A receives. Follow D3 before D2, the pending row, predecessor commit and D3 redelivery. | Actual source-event identity, before/after revision/count and receipt state. |
| 75–90 | A explains pending versus applied and challenges another repeat. | A's own action and receiving decision; unresolved recovery question and owner. |

These are independent lab contracts. The event-order runner does not authenticate webhook messages. The webhook runner does not establish source revision order. A valid MAC is not new business work; an applied local receipt is not real downstream business completion.

## Retain each person's work

Download the planner's blank individual worksheet once per participant. Each person fills their own copy with prediction, actual file/runtime/command, observation, contribution, assistance, changed-input evidence and receiving feedback. Save the editable agenda JSON separately. Share the public prepared-session link to share current authored context; it does not carry local notes or completed participant evidence.

Small groups can combine roles and say which check lacked an independent observer. Solo practice is self-review. The source/application owner proposes the next bounded raw-body/SDK, durable receipt, missing-predecessor or acknowledgement test; assigning a role does not grant real source or deployment authority.

For a separate client-request and lead-decision exercise, use the [sixty-minute pod rehearsal](https://training.lockedinlabs.ai/training/labs/pod-delivery) and its facilitator pack. Practice completion does not establish independence, qualification or placement.