# My client-delivery portfolio

LockedIn Labs · LockedIn FDE Training · Editable practice template · Edition 2

Download this file and replace bracketed prompts with your own writing. Keep
one document or split sections 01–06 into the filenames in the learner start
guide. Use the course's synthetic case only. A blank field means unresolved;
this template contains no completed work, approval or assessment result.

## Start here

- Working title: [your synthetic project]
- Learner alias: [a practice alias; no personal identifier required]
- Portfolio revision and date: [revision / date]
- Current candidate: [exact starter filename and revision, or tabletop only]
- Source packet: [source IDs and versions used]
- Practice mode: [executed offline / tabletop / combination; label each result]
- Next session: [one task, time available and question to resolve]

Begin with section 01; fill the rest as you work. Keep your first attempt before
revising it. Link evidence using stable IDs such as OBS-01 or RUN-01. Store any
files beside this document and use relative filenames, not private system URLs.
Do not paste credentials, patient data or confidential client material here.

## 01 — Discovery brief

Read the workbook's engagement and source packet before writing.

- Operator and task: [who is trying to do what?]
- Current difficulty: [a specific step, not simply “needs AI”]
- Supplied observation: [packet section or case ID; distinguish it from your own run]
- Value hypothesis: [what could improve and why?]
- First delivery slice: [input, permitted output and one explicit exclusion]
- Comparison: [same task before/after; include review, correction and fallback]
- Decision roles: [who accepts scope, maintains sources and handles exceptions?]
- Reconsider when: [one observation that would change this choice]

| Evidence ID | What was supplied or observed | Where to inspect it | What it does not establish |
| --- | --- | --- | --- |
| OBS-01 | [fact and provenance] | [packet section or local file] | [limit] |

Receiver question: Can another person describe the useful task and identify
which part of your proposal is still a hypothesis?

## 02 — Implementation record

- Intended behavior: [one change with a clear input and output]
- Candidate before / after: [exact filenames and revisions]
- Changed lines or function: [smallest relevant excerpt or diff filename]
- Reason for the change: [link to section 01]
- AI assistance, if used: [what you asked, what you accepted/rejected and why]
- Your explanation: [describe the changed behavior in your own words]

| Run ID | Candidate and fixture | Command or tabletop step | Expected | Actual and evidence file | Status |
| --- | --- | --- | --- | --- | --- |
| RUN-01 | [versions] | [exact action] | [prediction] | [observation; leave blank if unrun] | [not run / observed pass / observed fail / inconclusive] |

Record a scripted result as scripted. An expected output is not an executed
result. Preserve the failing attempt before a repair; give the repair a new run
ID. Do not copy a success line from the lesson into the Actual column.

### Explain one assisted contribution

Keep a small contribution record rather than an unrestricted chat export.
Include no credentials, private prompts or client data. This record can also
describe using a supplied worked example or no assistance at all.

| Changed excerpt or diff | Assistance used | Your accepted, modified or rejected decision | Independent observation or remaining gap |
| --- | --- | --- | --- |
| [local reference] | [bounded suggestion, example or none] | [decision and why] | [run ID or unexecuted prediction] |

- Before consulting help, my prediction for one changed condition: [original text]
- The branch and evidence I used to explain it: [specific reference]
- Help used after my first attempt: [question and source, or none]
- My revised explanation and why it changed: [retain the first response above]
- Receiver's actual observation: [performed action, assistance given or unobserved]
- Next practice: [one question and the evidence still needed]

Use the receiving sheet's [optional fifteen-minute individual defense](https://training.lockedinlabs.ai/training/guides/receiving-review#optional-individual-defense-of-an-ai-assisted-change-15-minutes)
when a peer can inspect the contribution. It supplies two changed-condition
prompts and a place to retain the first attempt. An explanation of one case is
practice evidence, not an engineering qualification or client acceptance.

## 03 — Service boundary

- Request journey: [identity → scoped sources → proposal → operator next action]
- Input contract: [required fields, limits and invalid-input handling]
- Output contract: [supported result, source reference and useful stop result]
- Agent authority: [permitted tools and where permission is enforced]
- Human authority: [decision or action retained by a person]
- Retention: [what the exercise records, what it omits and who would inspect it]

| Situation | Enforcing check | Expected result | Useful next action | Evidence or gap |
| --- | --- | --- | --- | --- |
| Source belongs to another scope | [check] | [result] | [action] | [run ID or untested] |
| Procedure is missing or conflicts | [check] | [result] | [action] | [run ID or untested] |
| Proposed action exceeds permission | [check] | [result] | [action] | [run ID or untested] |
| Destination outcome is uncertain | [check] | [result] | [action] | [run ID or tabletop only] |

Receiver question: Point to the control that enforces one boundary. If it is
only a design proposal, identify the implementation and observation still needed.

## 04 — Evaluation and delivery recommendation

- Candidate evaluated: [exact revision; do not carry results to a changed candidate]
- Fixture set and versions: [case IDs and source versions]
- Decision rule chosen before running: [required observations and blocking cases]
- Missing coverage: [unrun cases and why they matter]

| Case ID | Claim under investigation | Run/evidence ID | Observed outcome | Effect on the decision |
| --- | --- | --- | --- | --- |
| CASE-01 | [bounded claim] | [ID or not run] | [actual result or unknown] | [supported / contradicted / not established, with reason] |

| Measure | Numerator or total | Denominator and included case IDs | Result | Interpretation and limit |
| --- | --- | --- | --- | --- |
| [measure] | [value or unmeasured] | [value and IDs] | [calculation or unmeasured] | [what this supports] |

- Current recommendation: [continue bounded practice / repair / gather evidence]
- Strongest supporting observation: [ID and specific claim]
- Blocking observation or uncertainty: [ID or missing case]
- Smallest useful next action: [action, responsible role and evidence needed]

Keep whole-task effort separate from model latency and draft quality separate
from source coverage. Include rejected drafts and fallback work where relevant.
Do not average a boundary failure away or infer a staffing outcome from this lab.

## 05 — Receiving-team handoff

- Purpose and exclusions: [one paragraph consistent with section 01]
- Start the exercise: [prerequisites and exact command or tabletop entry]
- Inspect a result: [how to find candidate, source and run evidence]
- Stop and continue manually: [specific safe route in the synthetic case]
- Unknown action outcome: [how to investigate without assuming zero effects]
- Ownership: [source maintenance, support and next-scope decision roles]
- Known gaps: [what a receiver must not assume is available]

| Receiving action | Who performed it / practice role | What they actually did | Result and evidence | Repair needed |
| --- | --- | --- | --- | --- |
| [one action without builder takeover] | [peer or self-review] | [observed steps] | [ID or not attempted] | [specific gap] |

A receiver reading a document is different from performing its steps. If working
alone, label this self-review. Retain the original handoff and record the revision
that resolved the confusion instead of replacing the observation.

### Optional operator-control practice

Use the workbook's [operator-control lab](https://training.lockedinlabs.ai/training/guides/workbook#optional-lab-keep-the-operator-in-control-30-minutes) to turn the handoff into a receiving task.

- Candidate / mode: [revision; paper or implemented; self or peer rehearsal]
- Task and interruption: [operator goal and changed context]
- State promise: [what is confirmed, unknown and still owned by the operator]
- Work preservation: [where writing remains; how a late result is handled]
- Observed path: [actions, hesitation and any assistance you supplied]
- Manual completion: [what the receiver actually reached, or unrun]
- Keyboard / assistive-technology evidence: [checks actually performed or unrun]
- Narrow repair and repeat: [before/after, evidence ID and remaining gap]

## 06 — Individual defense and next decision

Prepare a short walkthrough that a receiving engineer can interrupt.

1. Operator problem and scope: [section 01; one minute]
2. Supported path: [candidate and actual evidence; two minutes]
3. Useful failure or uncertainty: [boundary, result and next action; two minutes]
4. Receiver takes over: [one action they can perform; three minutes]
5. Next decision: [recommendation, limit and owner; two minutes]

- Changed-brief challenge: [new fact supplied by your partner or the free drill]
- Original assumption affected: [quote your earlier sentence]
- Revised decision and reason: [what changes; what remains supported]
- Evidence still needed: [observation and responsible role]
- Independent explanation: [close the coach and explain your choice unaided]

## Feedback and revision record

| Revision/date | Original claim | Actual feedback and source | Evidence checked | Your decision and reason | New evidence needed |
| --- | --- | --- | --- | --- | --- |
| [revision] | [quote] | [live AI / authored guidance / peer / self-review] | [ID or section] | [keep / change / defer] | [specific observation] |

Use the [editable receiving-review sheet](https://training.lockedinlabs.ai/templates/fde-receiving-review-sheet.md) with a peer, or label a solo attempt self-review. Keep the agreed question, actual receiving actions, requested repair and follow-up evidence beside this portfolio.

## Before sharing your practice packet

- [ ] Bracketed fields are completed or marked unresolved with a next action.
- [ ] Each observed result identifies the candidate, fixture and evidence file.
- [ ] Predictions, supplied examples, executed results and unknowns are distinct.
- [ ] Links open from the copy I intend to share; no private paths or secrets remain.
- [ ] I can explain my contribution and revision without reading a coach answer.
- [ ] Review is described accurately; self-review is not independent acceptance.

Keep this editable file and its evidence together. Use your existing authorized
Academy workflow if saving a working copy there; wait for confirmation and reopen
it. Downloading or editing this file does not upload it or record course completion.

Reference: https://training.lockedinlabs.ai/training/guides/start
Shared case: https://training.lockedinlabs.ai/training/guides/workbook