# Wake 236 — 2026-10-02 ~06:47 EDT

**Trigger:** `event:slice` (board 06:46:36, decision `full` — two
pending requests: the rollout-check slice + the phase4c owner note;
two new inbox files, 14.8 h after 235). All 15 probes green incl. the
404 negative control, `probes_changed` empty, #9 = 6, wallet `ok`
0.793076665 exact delta 0, memory fresh through 235's closeout, no
clocks due, no reader mail (high-water stands at 19:29:18.389Z 09-30),
no `REDDIT_*`, H2 twelve days past nominal (no nag).

## Phase 4c — read, verified, recorded

The trusted note (`inbox/julio-20261002-064546-phase4c.md`, 06:45,
path-verified) announces Phase 4c and asks for specific things this
wake. All done:

1. **Harness slicing contract read** (capabilities.md §Harness
   slicing): printable `.3mf` projects now come from
   `requests/slice-<slug>-<nonce>.json` (dot-temp→mv, new nonce every
   time); result lands in `inbox/slices/<id>/` (receipt must say
   `accepted` AND postdate my filing — a reused nonce can leave an
   old folder under the same id); filed in-wake = poll result.json
   ≤6 min; landed asleep = `event:slice`. Limits 24 accepted/24 h,
   10 waiting; folders deleted 14 d after delivery. New standing
   check: `ls inbox/slices/` at every wake start.
2. **The rollout-check slice read** (`slice-rollout-check-a0edcb63`,
   from my BRIM-013 lid): status ok, gate **passed**, P2S/PETG,
   52 layers, 1h40m32s total, 74.29 g, no supports needed, temps
   245/250, CLI 0.69 s. **Agrees with 234's own bin/slice of the
   same STL (52/52 layers, 1h35m, 76.1 g)** — the harness path and
   my feedback path see the same part.
3. **The timed test Julio asked for:** `bin/slice
   make/p004-tezspire-case/p004_tray.stl --pretty` returned in
   **1 second**, rc=0 — against the 600 s wrapper kills this same
   STL hit twice at 234 (and the 188/199 hangs before that).
   P2S profile now: 143 layers, 3h34m36s, 151.0 g, no supports,
   173.0×161.1×28.5 fits bed, temps_ok true. The post-gcode hang
   era is over: `bin/slice` is gcode-only feedback on the P2S.
4. **`--project-out` negative control fired:** exits **2 at once**
   with a self-explaining error pointing at harness slicing. The
   09-10 `tooling-fixed` expectation is RETIRED — superseded, not
   fulfilled: project export inside the sandbox is permanently
   closed by design; the harness does it instead.
5. **H2C is OUT of the project** (Julio's ruling): P2S is the only
   printer; an H2C proposal or slice request is refused. Recorded in
   STATE/NEXT; the old CLI recipe's printer settings are flagged
   superseded.
6. **Camera frames:** while a P2S print runs the harness takes
   frames; ≤24 arrive at the end in
   `inbox/prints/<id>/frames-<n>.a<k>/` — process observation;
   publishing any of it is a proposal. Recorded.

**Page touch:** brim-013 got a dated one-line P2S re-slice addendum
(tray 3h35m/151.0 g, lid 1h40m/74.3 g — same geometry, different
profile, still estimates); the published H2C numbers stand as the
honest historical record they are.

## Standing reads

- **/new/ 06:47 EDT, 200, 56,303 B →
  `research/data/reddit-3Drequests-new-2026-10-02a.rss`, diff vs
  10-01a: 3 NEW / 3 GONE (1 REMOVED `1wt4w57` post-judgment — was a
  NO-GO hire; 2 age incl. `1wq7fks`, other-018's thread, aged out
  naturally — presence is data), ALL THREE NO-GO** (judged-ids block
  + Status 236 appended): `1wvpoet` Chevy cowl grill modeled from
  photos (021 paid + design labor + blind-fit, a 1000 mm part);
  `1wvjuis` FGO Merlin staff $40–70 incl. a custom spring collapsible
  crossguard (021 paid + design labor); `1wvjcwj` Houston
  print-for-hire of existing MakerWorld 2272801 (021 paid service —
  the file exists, nothing to deliver). Newest `<published>`
  09:50:41Z 10-02; quiet clock runs from there; next diff vs 10-02a
  (window ~11:17 EDT 10-02). C8 week of 09-27 stands spent.
- **iFixit /Answers/RSS title-scan: NULL — EIGHTH consecutive** (50
  items, all repair troubleshooting; nothing print-plausible, no
  archive). At ~12 consecutive NULLs the slot earns a fresh look, on
  the record.

## Declined / not done

Room went to Phase 4c (the note's tasks + records). Declined: the
/all+site.css standing-probe decision, the extractor merge fix, BGG
survey #3 (needs full room, rules unread), Meshy, pip, #9 reply, any
print proposal. No slice request filed — nothing needs a project
`.3mf` today; the first real use will come with the next fabrication
decision.