Fills the mood slot FELT_STATE_ARCHITECTURE deferred on 2026-05-15.
A belief ladder under both cadences, reusing the device_state doctrine:
actors are inputs, not authorities. Four types kept apart — exudate,
belief, impulse, intention — so mood and atmosphere cannot reconflate.
TASK-36 said the load path had never executed. It ran 2026-08-11 —
LoadNanoappResponse decoded, success byte set, and six gauges came off a flat
baseline at 2502 and 2990 against a 2500 floor. Recorded with Casey's numbers
and the two things that would otherwise compress away: the margin carries
intent, and the proximity veto is a separate job from the threshold.
TASK-66 corrected. The shipped front page promoted ViewTop to a peer of the
substrate, the OS and the agent; the canonical map never agreed. Three layers
now, plus the delivery defects found by auditing served bytes.
Reconciles the two owning docs with souveraine cfad688. Sub-scope A was
always half-landed: the substrate was multimodal end to end and the HTTP
door was text-only, so the TUI could see and every other surface was
blind. The door opens; nothing walks through it yet.
TASK-72 keeps image intake — picker, chip, base64, rendering — and now
has a server to call.
§2 closed, §3 built and half-accepted: the split reaches the message and
serialises, but no test asserts it and no surface renders it. §1's config
change is recorded as mine, not Casey's word. Run 1283 green, r399 shipped
to edge and running on the laptop; the phone has not pulled it.
The config hands the subconscious sliding_window where the doc and the
code default both say sliding_reflect, and the memory tool exposed the
blind strategy while hiding the safe one.
Also indexes 63, 64 and 66 — all three existed on disk and in no table.
Measured once: reply commits and PrimaryComplete fires, then the N+1
pass stops at round 3 of 10 with a fully idle runtime. The pass runs
inside run_turn, so tx never drops and the SSE stream never closes.
Lead is the todo tool. Stream lifetime is a separate fix.
Display pipe binds and the panel stays dark; touch worked once and stopped.
Both have a named single-boot test rather than a theory. Driver changes and
rails are ruled out by test; the display path is not.
blueline-screen-toggle does not exist here so the blank no-ops, and Dim is
an absolute that reads as off on a 2047-step backlight. Lit-forever or
black-forever, nothing between, until eb0292c ships.
One row per capability, one column per body, with a strict status vocabulary —
"works" means exercised with evidence named, never that a device node appeared.
blueline points at PAF/ rather than copying it; d10 carries its own state; m1
records what already transfers before anything is measured.
The display pipe is corrected in the same change: ADP binds and scans out and
the panel stays dark, so it is partial, not working.
Touch proven with real events, not a device node: three layers, the last
being runtime report activation gated to D11 so D10 went silent a second
after boot.
adp binds on 206200000.display-pipe with card1-DSI-1 — the first DRM
display pipe on this hardware; everyone else scans a dumb framebuffer.
Still unused because m1n1 re-enables /chosen/framebuffer and udev then
hands viewtop simpledrm.
Kernel patch kept as evidence; hoolock/ is not a repo.
The shell is on the iPhone 7 — viewtop, sessiond and qs, drawn by llvmpipe
on a card with no render node, over one cable. Wallpaper, palette, bar,
clock, dock and her quote match the Pixel.
Everything that blocked it was the base assuming blueline. That rootfs had
never installed a package in its life and could not have: no Landlock, and
no keyring at all. The phantom pacman options TASK-25 caught in July had
propagated into a second image. PAM, the panel-blank script and the fonts
all had to be hand-carried because they ship by flash or not at all.
The lock is the good news. Root on the box and a key in the account, and
viewtop, sessiond and the shell each refused the unlock in turn.
fprintd, gazed, sailfish-fpd and biometryd are the same interface four
times. Take that shape and face and finger become one surface with two
backends. Mapped every call onto the recovered TA commands; the gaps
are persistence and the auth token, both already named.
Prior art checked: sailfish-fpd-community needs hybris, wrobelda's
libfprint branch is the only match-on-chip-over-QSEE example, and
nobody has done FPC over QSEE natively. One listener receiver per
device means the supplicant is a separate owner from the fp daemon.
The awkward hand-off was not the dwell gate — it was committing at whatever
shift the thumb left. quickstep travels to the end target first; settleTo
does that now. Dwell restored to match prior art.
Records the geometry being measured rather than assumed, and two findings
that are not this task's: furniture_at() is a stub so synthetic touch cannot
reach the pill, and Origin is two-valued so agent touches carry no audit.
Two writers own one geometry — the rail's pose and ZoneOverview's scale,
running opposite curves off the same progress. Three attempts to make the
numbers agree failed in a session. The strand proves it: the repro path
is card taps, which never reach the rail's clear-the-pose code.
Plan follows quickstep, which is on this machine: one shift 0..1, the
live window transformed onto the real card rect, end targets as an enum
with LAST_TASK a destination rather than an abandon branch, drag length
derived from that rect.
The transition is a verb because she gives the tour, under the origin
rules input.rs already enforces: agent input is not evidence and the
lock surface takes fingers only.
