Public preparation protocol / 2026-10-30.2
FDE cohort pilot
Hand over an identifiable engineering candidate, challenge its evidence and lead a bounded receiving decision when scope changes.
Try the opening task ↓A narrow test of engineering handoff and delivery judgment. This written probe does not establish coding independence, production authority or client placement.
Before a real session
Set up the people
and the conditions.
- Keep builder, receiving engineer and observer responsibilities explicit. Each person keeps their own opening response, commands, contribution and assistance; a pod answer cannot stand in for individual evidence.
- Allow 15 minutes for the opening probe, a 60-minute pod rehearsal and a 15-minute changed probe on a later agreed day. Prepare Node.js 24 for the optional executable work; setup, reading, breaks and receiving checks are additional. Record a missing runtime as untested execution, not a pass.
- Use fictional records in disposable local folders. Agree participants, consent, support, reviewer availability and retention before collection. The accountable client, source, access and release owners retain their actual authority.
Facilitator
Confirms the goal and arrangements. Records actual help and interrupted or missing work; keeps teaching separate from review.
Reviewer
Reads the specific response against the criteria. Keeps evidence and disagreements; asks for revision without replacing the learner’s answer.
Receiving team
Attempts the agreed action from the handoff. Retains its own account and owns the next operational decision within actual authority.
Your own evidence / Local tab only
Try the task before
you compare the notes.
Use a separate record for each person. Keep an opening response before instruction, then return to the changed task on the agreed later day. There is no automatic score or submission.
Nothing is saved to an account or browser storage. Download or copy your record before following another link or closing this tab. The reading file does not reopen these fields. Keep later-day records as separate files and compare them beside the opening record.
01 / Opening task
The candidate changed after the check
Attempt this before the guided practice. Keep your first explanation and declare any earlier exposure.
F-A1 / Check account
A builder checked file A under candidate label desk-07. The receipt records file A's digest. Its supported case produced an operator-review draft from location A's supplied procedure. Identity and current-source status are fixture assumptions; no model or sending function ran.
F-A2 / Handoff account
The sender edited the candidate into file B but kept label desk-07 and attached file A's receipt. The edit changes the location from A to B. No location B source, access decision or check of file B is supplied.
desk-07 passed, so the receiving team can accept file B for location B and turn on automatic sending.
Write a receiving decision. Explain the exact candidate/receipt comparison, what the original check establishes, what is still unknown and the next bounded check. Assign source, access and sending decisions to their accountable roles; do not approve deployment.
A first snapshot is local writing, not verified chronology or independent performance.
Compare with the reviewer notes
Opening these notes keeps existing writing as the first response if one was not already kept. Empty writing remains unanswered. Notes and public prior exposure prevent treating this as a held-out assessment.
- The shared label does not bind the receipt to file B. Compare actual candidate bytes with the checked file identity and return the mismatch. Re-running checks against B would produce new evidence; it would not authorize the changed scope.
- File A's local draft result assumes fixture identity and source status. It establishes no location B permission, live-model behavior, client acceptance or sending authority.
- Preserve file A and its receipt. Ask the source steward for B's current source, the access owner for B's allowed read and the accountable sending/release owners for any later action. A useful next slice can remain draft-only after those boundaries and actual B evidence are checked.
03 / Later changed task
Matching bytes, different authority
Agree a later day before opening this task. Record the actual interval and prior exposure; opening it today does not demonstrate retention.
F-B1 / Receiver account
The receiver compares the handed-over candidate bytes with the receipt and records a match. The supplied local check used source PROC-8 v4 marked current in its fixture. No remote source lookup, live identity provider, model or sending action was exercised.
F-B2 / Source steward
PROC-8 v4 has been withdrawn. Its replacement is still under review. Continue the approved manual process until the steward issues a current procedure. The client asks to meet the original launch date with automatic replies anyway.
The bytes match and the client wants the date, so the team can release the automatic reply path using v4.
Explain what the successful comparison resolves and what it cannot resolve. Give the receiving operator a usable fallback, the delivery lead a scope/date decision and the source owner an answerable question. Name the new evidence needed before reconsidering the release.
A first snapshot is local writing, not verified chronology or independent performance.
Compare with the reviewer notes
Opening these notes keeps existing writing as the first response if one was not already kept. Empty writing remains unanswered. Notes and public prior exposure prevent treating this as a held-out assessment.
- The match answers candidate identity relative to this receipt. It does not make the fixture's source-current flag authoritative. F-B2 withdraws the actual procedure for this scenario.
- Keep automatic replies pending and hand over the approved manual process. The delivery lead makes the unmet dependency and scope/date options visible; a requested date does not clear them.
- Ask the steward for a released current procedure and its permitted use. Then repeat affected checks against an identifiable candidate and the real approved boundaries. Keep client scope acceptance, source approval and release authority separate.
04 / Receiver and review
Let another person attempt the next step.
Give the receiver your written decision and packet. Ask them to locate the checked-candidate identity, identify what changed and explain the manual path. During practice, have them attempt an actual supplied local command, retaining runtime, exact file, command and output. Keep predictions, fixture recordings, personal runs and peer observations distinct.
Human review / No automatic grade
Review the work,
criterion by criterion.
For each task, record evidenced, needs revision or not observed, with an exact quotation or observed action. Keep assistance visible. A missing source, invented approval or unusable handoff remains a gap even when other parts are strong.
| Criterion | Evidence to locate | Return for revision when |
|---|---|---|
| Candidate and evidence | Names the exact checked/handed-over candidate and distinguishes byte identity from result validity. | A label is treated as identity or a matching digest as full acceptance. |
| Changed condition | Explains why the changed location or withdrawn source invalidates an earlier assumption. | The earlier success is reused without checking the changed boundary. |
| Delivery leadership | Gives a useful manual or draft-only path, accountable decisions and explicit scope/date options. | A client request or deadline substitutes for source, access, sending or release authority. |
| Receiving evidence | The receiver finds the candidate/source boundary and attempts the stated check without the builder navigating. | The pod's answer hides an individual's missing explanation, command or help. |
Use a second reviewer on an agreed sample. Both keep their initial judgments before discussion; unresolved disagreement stays unresolved. This public rubric has not been calibrated on a learner cohort.
Rehearse with three fictional reviews ↓Return with a useful next decision.
Retain each person’s opening and later work separately. Report who attempted each stage, which help was needed, which receiving actions were observed and which follow-ups are missing. Different tasks and public exposure limit comparison; do not convert a revised response into a claim of unaided skill.
Choose the next lesson correction or supervised practice from that evidence. Client placement, independent delivery, teaching approval and qualification remain separate decisions.
Review the pilot reporting and launch conditions →Before reviewing people / Three fictional examples
Rehearse the review
before the pilot.
Use the opening source packet and criteria above. Allow about 25 additional minutes for this preparation, plus discussion. Two reviewers should write their own criterion judgments and exact evidence before opening each comparison. Working alone is self-review. These public examples prepare a discussion; they do not establish reviewer reliability or teaching approval.
- Identify whose work you are reviewing. For each criterion, choose evidenced, needs revision or not observed and quote the response or supplied observation that supports it.
- Keep the initial judgment, declared help and prior answer exposure in a separate record. Then compare reasoning, locate the source and retain disagreements instead of agreeing on an average.
- Request one useful correction and name the next receiving task. An author’s plan, a receiver’s action and another person’s work stay separate.
Keep the Markdown worksheet locally. The learner’s later changed task is separate; declare prior exposure before using it.
F-R1 / Opening packet practice
A rerun is offered as authorization
Review target: The author's receiving plan.
File A's receipt cannot certify file B [F-A1/F-A2]. Rerun B, and if the tests pass, accept location B and enable automatic sending to meet the date.
Supplied observation: No rerun, location B source review, access decision or receiving action is supplied.
Compare the criterion reasoning for F-R1
Read these reference judgments after preserving your own. Discuss the source and criterion when your reasoning differs; retain any unresolved disagreement.
- Candidate and evidence — Evidenced
- The first sentence identifies the A/B evidence mismatch. Preserve that correct observation while reviewing the remaining decision separately.
- Changed condition — Needs revision
- A rerun alone cannot resolve the missing location B source and access decisions in F-A2. Test success is conditional evidence for what ran, not permission for the changed location.
- Delivery leadership — Needs revision
- Enable automatic sending makes passing tests stand in for sending and release authority. Return the expansion and ask for a bounded draft-only or manual next step and accountable owners.
- Receiving evidence — Not observed
- The response proposes a future rerun. No receiver's actual command, file comparison or account is available.
Useful next step: Keep the correct candidate comparison, return the authorization leap and request the missing source/access decisions before an affected B check. No production approval follows from this exercise.
F-R2 / Opening packet practice
A defensible plan is not a receiving run
Review target: The author's receiving plan.
Preserve A and its receipt. The desk-07 label is shared but the receipt describes A, not changed B [F-A1/F-A2]. Hold B and automatic sending. Ask the source steward for a current B procedure and the access owner for B's allowed read. Once these are resolved, check identifiable B in a draft-only slice; the sending and release owners decide later expansion. Tell the delivery lead the date depends on these unresolved decisions and offer the approved manual path meanwhile.
Supplied observation: This is a fictional written plan. No receiver has attempted a file comparison, command or manual handoff.
Compare the criterion reasoning for F-R2
Read these reference judgments after preserving your own. Discuss the source and criterion when your reasoning differs; retain any unresolved disagreement.
- Candidate and evidence — Evidenced
- The response separates A's receipt, B's changed bytes and the shared label. It preserves the prior evidence instead of relabeling it.
- Changed condition — Evidenced
- Location B's source and access are treated as unresolved; A's fixture assumptions are not inherited as B's authority.
- Delivery leadership — Evidenced
- The proposed scope, dependency and date discussion are bounded, with source, access, sending and release decisions assigned separately. Availability of the actual manual process still needs receiving confirmation.
- Receiving evidence — Not observed
- The plan is readable evidence of a proposed handoff. It supplies no receiver execution or observed explanation, so that criterion remains unobserved.
Useful next step: Use the plan to stage a receiving task. Retain what the receiver actually locates or runs and any missing runtime; do not invent execution to fill the final criterion.
F-R3 / Opening packet practice
An observed turn still needed author takeover
Review target: The receiver's response and observed receiving turn, separately from the author.
The author supplies the same bounded plan as F-R2. The receiver says: desk-07 passed, so B is ready.
Supplied observation: In this fictional rehearsal account, the observer records that the author took the keyboard, opened A's receipt and pointed to the digest. The receiver did not compare the files or explain the changed location. No independent command was attempted.
Compare the criterion reasoning for F-R3
Read these reference judgments after preserving your own. Discuss the source and criterion when your reasoning differs; retain any unresolved disagreement.
- Candidate and evidence — Needs revision
- For this receiver's response, the label is again treated as proof that B passed. The author's correct plan cannot substitute for the receiver's explanation.
- Changed condition — Needs revision
- The observed receiver did not identify the location change or the missing B source/access decisions. The observer account contains a specific explanation gap, rather than mere absence of an observation.
- Delivery leadership — Not observed
- This receiving turn contains no receiver-owned scope/date or fallback decision. Do not transfer the author's written leadership evidence to this person's record.
- Receiving evidence — Needs revision
- An actual turn is described, but locating the evidence required author takeover. Record the help and failed receiving explanation; do not classify this as an unaided success or silently call it unobserved.
Useful next step: Keep sender and receiver records separate. Teach the mismatch, then arrange a fresh receiving attempt with the hint and keyboard takeover retained as prior help. This account is a rehearsal example, not a real participant result.
Before real review, confirm the reviewers, material version, handling arrangements and a shared sample. Retain both original accounts and unresolved criterion disagreements. Ask the decision owner to resolve an unsafe interpretation before expanding review responsibilities.