results and products; add epilogue prompt for this session

# Day 5 Summary Prompt — for a fresh Claude session

*Save this. Open a new Claude session, paste the prompt below, and

attach the indicated files. The prompt is self-contained — Claude

will not have read prior summaries unless you direct it to.*

---

## Prompt to paste

> You are working in the `tasking` project, `hack-a-thon-2026`

> branch. This is the **Day 5** (closeout day) summary of a

> five-day Reflectometry hack-a-thon. Days 1–4 are already

> summarized at:

>

> - `tasking/Hack-A-Thon-2026-Day1-Summary.md`

> - `tasking/Hack-A-Thon-2026-Day2-Summary.md`

> - `tasking/Hack-A-Thon-2026-Day3-Summary.md`

> - `tasking/Hack-A-Thon-2026-Day4-Summary.md`

>

> Plus an estimation baseline at:

>

> - `tasking/Hack-A-Thon-2026-EWM-Estimation-Baseline.md`

>

> **Read those five files first** so you understand the strategic

> ordering, the three critical scientific issues, the Working Plan

> consensus from Day 4, the MVP architecture direction (`research_mvp_architecture.md`),

> and the EWM-derived effort baseline (358 h median / 816 h p80 over 39 backlog items).

>

> The Day 4 summary's §16 lists the explicit non-decisions and

> §17's "Day-5 action items" are the going-into-Day-5 commitments

> — these are what Day 5 should resolve or carry.

>

> ---

>

> ### Day 5 source files (in `~/Desktop/Hack-A-Thon-2026/`)

>

> Bogdan has handwritten notes and transcripts. **Always read the

> manifest first** (it tells you what files exist and which PDF

> pages are scanner-inverted):

>

> - `2026-04-28-Files.md` — manifest (read this first)

> - `2026-04-28-Part1-transcript.txt` — Teams transcript, AM

> - `2026-04-28-Part2-transcript.txt` — Teams transcript, PM (if any)

> - `<scanned-pdf>.pdf` — Bogdan's handwritten notes (the manifest

>   will list the actual filename and any inverted pages)

> - `QuickNXS-Hackathon-2026-GoogleDrive-*/Notes 04-28-2026.docx`

>   — the live collaboratively edited Google Doc; **authoritative

>   for Day 5 outcomes** (treat the same as Days 2–4).

> - Possibly an updated `Notes 04-27-2026.docx` if the team

>   retrospectively edited Day 4 notes — diff against the Day 4

>   .md mirror Bogdan saved (he'll provide a diff file

>   `~/Desktop/Hack-A-Thon-2026/2026-04-27-Notes-GoogleDoc-diffs.md`

>   if any edits occurred).

>

> ---

>

> ### Per-day workflow (proven across Days 1–4)

>

> 1. **Read the manifest first** to enumerate sources and find

>    inverted-page flags.

> 2. **Read the Google Doc `.docx` first** — it's the authoritative

>    record of decisions; everything else is corroboration.

> 3. **Read transcripts second** (small, fast; useful for

>    quote-resolution). Teams auto-transcripts are noisy — speaker

>    attribution is unreliable; product names get garbled

>    (`mr_reduction` → "demanded Albertson"; `refl1d` → "refer 1D";

>    `LR_reduction` → "LRRed" or "referend").

> 4. **Render the PDFs:**

>    `mkdir -p /tmp/d5 && pdftoppm -r 150 -png <path>.pdf /tmp/d5/p`

>    Then check each page; many pages are scanner-inverted even

>    when the manifest doesn't flag them. Use

>    `sips -r 180 /tmp/d5/p-XX.png` to rotate. Verify by reading.

> 5. **Read every PDF page in full** — the handwritten notes are

>    the most authoritative record of who-said-what after the

>    Google Doc.

>

> ---

>

> ### Day 5 agenda anchors (from the kickoff agenda doc)

>

> - **09:30 – 09:45** — Summary of Day 4 and today's charge (Bogdan)

> - **10:00 – 12:00** — Session 10

> - **12:00 – 13:00** — Lunch (Salsaritas)

> - **13:00 – 15:00** — Session 9 *(yes, the agenda says 9 — likely a

>   typo for 11; corroborate from the day's actual schedule)*

> - **15:00 – 15:15** — Break

> - **15:15 – 16:00** — Final Document Preparation

> - **16:00 – 16:30** — Closeout (Bogdan)

>

> **Threshold Goal:** Triage and Estimation of the EWM tickets

> generated by COB Day 4 *(the ticket-generation that was the

> Day 4 stretch — confirm whether tickets were actually generated

> and what their status is by morning of Day 5)*.

>

> **Stretch Goal:** Completed separation of QuickNXSv2 into its

> Qt-based front-end and the MR reduction back-end, verified and

> tested with all tests passing. *(This is ambitious for one day;

> partial completion is the realistic outcome.)*

>

> ---

>

> ### Things to call out specifically in the Day 5 summary

>

> - **Did the three-step strategic ordering survive?** (1) restructure

>   QuickNXS BE/FE first; (2) work on LR new BE; (3) integrate LR

>   into QuickNXS BE. Day 4 was unanimous; Day 5 might revise.

> - **What happened with the three critical scientific issues?**

>   Off-spec output (Asmaa+Valeria), run summing (Becky/QA), off-spec

>   binning (Marie hide in UI). Did Valeria's V1 baseline arrive?

> - **Did the team produce technical-debt stories?** This was the

>   end-of-Day-4 "if we get through stories of technical debt, we're

>   in a pretty good spot" target. Map any new stories back to the

>   Day 3 39-item backlog and the EWM Estimation Baseline.

> - **Did the team produce EWM tickets?** Threshold goal — get the

>   Id, Type, Owner, Estimate, Planned For for any new tickets.

>   Consider re-running `/tmp/phase1_extract.py` against the latest

>   EWM xlsx if Bogdan provides one (or note that the baseline

>   needs refresh).

> - **MVP architecture progress:** did the team start touching

>   `MainHandler` / introduce `IMainView` / replace

>   `auto_change_active`? `research_mvp_architecture.md` is the

>   reference; report which Phase (1–6) was actually executed.

> - **Settings.json and the Configuration class split:** did either

>   land as code, or only as design?

> - **`refl1d` on NDIP:** did Bogdan kick off the 0.11 → 1.01

>   update task? (Day 4 he committed; Day 5 may be progress.)

> - **Closeout commitments:** at the 16:00 closeout block, what

>   did Bogdan articulate as the final dispositions? *(This is the

>   most important content of the day — the final-document-prep

>   slot at 15:15 is where the team consolidates; the closeout is

>   where Bogdan summarizes what got committed.)*

> - **Date for the second Hack-A-Thon** (committed Day 2 to before

>   the 2026-06-25 → 2026-08-04 SNS beam outage) — does Day 5

>   pick a target week?

> - **Code-deep-dive with Becky** (committed Day 4: 2-day session

>   on Becky's `lr_reduction` new-workflow code, with Bogdan/Becky

>   to harden the code first) — was this date pinned?

>

> ---

>

> ### Style / structural conventions for the summary

>

> Match Days 1–4. Each summary has:

>

> 1. Header with location, hours, threshold/stretch goals, outcome

>    (✅ or partial), and consensus headline.

> 2. Sources block listing every file used.

> 3. Attendees reconstructed from notes.

> 4. Per-agenda-block sections.

> 5. **Consensus & decisions** numbered list — these are the

>    citable outputs.

> 6. **Explicit non-decisions** — what got deferred.

> 7. **Open questions / Day-N+1 charge** — what carries forward

>    (or if Day 5 is the closeout, what carries forward to the

>    deep-dive / Hack-A-Thon #2).

> 8. **Notes on fidelity** — caveats and ambiguous tokens.

>

> Target length ≈400–600 lines of Markdown; Day-1–4 summaries

> ranged 220–650 lines depending on density.

>

> ---

>

> ### Writing the prompt-for-the-next-session

>

> Day 5 is the closeout, so there is no next session. **Instead,

> at the end of the Day 5 summary, produce a "going forward"

> section** that captures:

>

> - The deep-dive scheduling (date, attendees, scope)

> - The Hack-A-Thon #2 scheduling (date, scope)

> - Open EWM tickets generated this week, with owners

> - The next concrete code-level milestone (probably:

>   "introduce `IMainView` Protocol and replace `auto_change_active`

>   with `blockSignals()`" — the lowest-risk first refactor PR)

> - Any commitments outside the team (NDIP `refl1d` update; Mantid

>   substrate working group from the optional Day-3 Mantid Future

>   meeting; ORNL-Assistant-on-moderate-data demonstrations from

>   the AI Lunch)

>

> ---

>

> ### Things you do NOT need to do

>

> - Do not re-read Days 1–4 PDFs or Google Docs. The .md summaries

>   are sufficient context.

> - Do not redo the EWM Estimation Baseline. The Day-4 baseline is

>   current; if Bogdan provides an updated xlsx, run the existing

>   script `/tmp/phase1_extract.py` (preserved in the Day-4

>   summary's reproducibility section).

> - Do not extend the architectural research. `research_mvp_architecture.md`

>   stays the working reference; Day 5 should only report what was

>   *executed* against it.

>

> ---

>

> ### Final deliverable

>

> Write the summary to `tasking/Hack-A-Thon-2026-Day5-Summary.md`.

>

> Watch your context. If you reach 75% utilization before

> finishing, write a checkpoint plan to

> `tasking/Hack-A-Thon-2026-Day5-Summary-PLAN.md` and let me

> review before continuing — same protocol as Day 4.

>

> Take your time. This is the closeout summary and is the most

> citation-worthy of the five — it's the document that

> stakeholders will read to understand what the hack-a-thon

> committed to.

---

## Notes for Bogdan (not part of the prompt)

- The Day 4 summary preserves the AI Lunch and Learn content in

  §12. That side-track is **not part of the hack-a-thon** but is

  contextually important — it formalises the AI-tooling

  authorization and demonstrates concrete agentic-engineering

  precedents the team can leverage. Day 5 may reference it; the

  Day 5 summary should not re-summarise it.

- The reproducibility scripts at `/tmp/phase*.py` from the EWM

  baseline are temporary. If you want them preserved for future

  use, copy them into the repo (e.g.

  `tasking/scripts/phase{1,2,3}_*.py`).

- The Day 4 summary's §10 cites the Copilot-authored

  `research_mvp_architecture.md` as a working artifact — note that

  this satisfies the **Day 4 stretch goal** (code review of agentic

  AI development artifacts) at least partially. Day 5 may want

  to articulate whether this constitutes "stretch goal met."

- The 2nd Hack-A-Thon is an open commitment with a hard deadline

  (before 2026-06-25). Pin a date during the Day-5 closeout if at

  all possible.

---

*Prompt prepared 2026-04-27 evening. Use it Monday/Tuesday after

Day 5 concludes. — Claude*

go skip docx-diffs. I have inspected the files and the only difference I saw, I collected into ~/Desktop/2026-04-24-Notes-GoogleDoc-diffs.md which you may read. When you finish, please produce a prompt that I can use in a fresh session to incorporate the upcoming Day 5 activities that will capture the most important aspects of this session. Thank you so much for this amazing work!

```

⏺ Done. Two deliverables in tasking/:                                

  1. Hack-A-Thon-2026-Day4-Summary.md (~440 lines, ~17 sections) — the most consequential summary of the week. Headline:               

    - Three-step strategic ordering — UNANIMOUS: restructure QuickNXS BE/FE → work LR new BE → integrate LR into QuickNXS BE.

    - Three critical scientific issues catalogued and assigned: off-spec output (Asmaa+Valeria, against V1 baseline), run summing      

  (Becky/QA verify, fix already in QA), off-spec binning (hide in UI for now).

    - Working Plan ratified verbatim from the live Google Doc; freeze main dev branches; isolate refactor branch.

    - MVP architecture adopted as target using Copilot's research_mvp_architecture.md (962 lines) — two-phase Mutation+Render replaces

  auto_change_active; 6-phase migration.

    - NDIP confirmed as deployment substrate (John Duggan's overview); Bogdan's refl1d 0.11→1.01 task scoped.

    - AI Lunch and Learn delivered — separate side track, formalises ORNL-internal AI authorization with concrete Wireshark/Galil

  precedents.

    - settings.json / Settings singleton elimination decided.

    - Configuration class split into Global/Per-Run/Plot ratified.

    - GenX templates removal confirmed.

    - "Ten levels of danger" electrical-wiring analogy captured (Bogdan's framing for refactor depth).

  2. Hack-A-Thon-2026-Day5-Prompt.md — self-contained prompt for a fresh session. It tells the next-session Claude:

    - To read the five existing summary files first.

    - The proven workflow (manifest → Google Doc → transcripts → render+rotate PDFs → write).

    - Day 5 agenda anchors and threshold/stretch goals.

    - Specific things to call out: survival of the 3-step ordering; critical-issue dispositions; whether tech-debt stories were

  produced; whether EWM tickets were generated (the Day-4 stretch / Day-5 threshold); MVP execution progress (which Phase landed);

  settings.json / Configuration split status; closeout commitments; 2nd Hack-A-Thon date; deep-dive scheduling.

    - Style conventions (matches the Days 1–4 layout).

    - The 75% context-checkpoint protocol that worked Day 4.

    - A final "going forward" section requirement (since Day 5 is the closeout — no Day 6).

  Plus: the temp render directories and intermediate transcripts are cleaned up. The reproducibility scripts at /tmp/phase*.py from the

   EWM baseline are noted in the Day-5 prompt as worth promoting into the repo if you want them preserved long-term — let me know and

  I'll do that as a single git mv-style copy.

  Best of luck with Day 5, Bogdan. The hack-a-thon is in good shape going into the closeout — the consensus from Day 4 is one of the

  strongest single-day outputs of the week, and the team has a clear path to start cutting code.

```

Claude produced [Hack-A-Thon-2026-Day4-Summary.md](../Hack-A-Thon-2026-Day4-Summary.md) and [Hack-A-Thon-2026-Day5-Prompt.md)(../Hack-A-Thon-2026-Day5-Prompt.md)

( *I plan to make edits to the Day 4 summary post-commit.* )

### Prompt 1.8 (Epilogue)

Thank you so much! Before we say goodbye, is there anything we should remember from this (long) sesssion that would be useful in the cross-project knowledge system?