1. Three numbers to watch

If these move the right way, the path is working. If they don't, no explanation should override the numbers.

1a. Share of work through the new system

Turns handled end-to-end by the new system (strict measure)

No data yet. no published strict end-to-end "turns through new code" metric artifact yet — the stricter full-path number is not generated by the current scoreboard

Honest substitute shown below (writes-metric). The strict number must climb toward 100% once the full-path metric is published.

69.1%
Turns whose write evidence points to the new system (this is the writes-metric, not the strict full-path measure)

1845 new-code writes · 75 old-code writes · 751 not measured · out of 2671 turns in cycle 65c84eeb-c58-20260723. Of turns where code path could be measured: 96.1% new.

Source: latest test-trace cycle, counting which code performed the write (writes-metric — not the stricter full-path metric)

1b. Old vs new pass rates (lines will cross)

Pass rates by which code handled the turn — the "lines will cross" view. New code should catch and then beat old code, category by category. Percents use only graded pass/fail rows in each code group; "no evidence" rows are excluded from the percent.

Test category Overall pass rate New-code pass rate Old-code pass rate Could not classify
Adversarial
w2h-wave3adv-grade.json
 31%
31 pass / 52 fail
 50% (5 pass / 5 fail) 50% (10 pass / 10 fail) 197
Clarifying questions & declines (conversational)
fanout-conv-w8-20260718-grade.json
 18%
33 pass / 13 fail
 no graded rows no graded rows 0 · 66 missing trace
Compound dishes
fanout-compound-dishes-A-fab27-20260718-grade.json
 0%
0 pass / 2 fail
 no graded rows no graded rows 2
Food A
fanout-foodA-grade.json
 0%
0 pass / 0 fail
 no graded rows no graded rows 0
Food B
fanout-foodB-w8-20260718-grade.json
 0%
0 pass / 0 fail
 no graded rows no graded rows 0
Food commands by type (log/update/delete/query)
fanout-foodtyped-w8-20260718-grade.json
 19%
19 pass / 20 fail
 no graded rows no graded rows 0 · 40 missing trace
Food mixed C (everyday)
food-corpus-mixed-C-65c84eeb-c58-20260723-grade.json
 43%
86 pass / 90 fail
 20.8% (16 pass / 61 fail) 20% (1 pass / 4 fail) 94
Food mixed D (everyday)
food-corpus-mixed-D-65c84eeb-c58-20260723-grade.json
 37%
73 pass / 106 fail
 21.1% (19 pass / 71 fail) 0% (0 pass / 4 fail) 85
Lists & settings (new-types)
fanout-newtypes-w8-20260718-grade.json
 13%
12 pass / 8 fail
 80% (12 pass / 3 fail) 0% (0 pass / 5 fail) 0
Multi-item meals
fanout-multifood-grade.json
 2%
1 pass / 34 fail
 0% (0 pass / 1 fail) 3.3% (1 pass / 29 fail) 4
Notes/todos/lists
fanout-notes-w8-20260718-grade.json
 19%
28 pass / 41 fail
 no graded rows no graded rows 0 · 73 missing trace
Query frozen (exact keys)
query-frozen-A-65c84eeb-c58-20260723-grade.json
 3%
2 pass / 10 fail
 16.7% (2 pass / 10 fail) no graded rows 0
Utility
fanout-util-w8-20260718-grade.json
 30%
13 pass / 6 fail
 76.5% (13 pass / 4 fail) no graded rows 2
Workout
fanout-workout-w8-20260718-grade.json
 2%
6 pass / 10 fail
 no graded rows no graded rows 16
Workout commands by type
fanout-workouttyped-w8-20260718-grade.json
 17%
17 pass / 1 fail
 94.4% (17 pass / 1 fail) no graded rows 0

1c. Instruction checks still green

6 of 6
Standing-instruction checks currently green

Recomputed from source at page build time. A red check that stays red past one cycle means the machinery is being overridden.

StatusCheckHoward's instruction (as written)
green — check passed Keep working on the new code first
new-code-first
 Keep working on the new code, make it better, the lines will cross. Old code is frozen reference only.
green — check passed Legacy main app file stays frozen
apptsx-frozen
 App.tsx is radioactive: nobody touches it ever again, for any reason, except its eventual deletion.
green — check passed Only Howard dismisses defects
only-howard-dismisses
 Only I dismiss defects.
green — check passed Cardio stays in the backlog
cardio-backlogged
 I don't want to focus on cardio logging right now; it belongs in the backlog.
green — check passed Defaults and history always on
defaults-history-always-on
 We definitely always want to use defaults and history — can't imagine when we want that off.
green — check passed Direction changes must be written down
direction-changes-surface
 Nothing gets decided against my instructions without telling me. (2026-07-21)

2. Decision log

Headlines from the daily session book (today and yesterday). Each line is a real decision or discovery recorded the same hour it happened.

2026-07-23 from docs/reset/For Other Codex/SESSION_BOOK_20260723.md

  1. ~1:15am ET — MEASUREMENT RESTORED: build 35679d4d = module-live + prod + TELEMETRY credentialed
  2. ~1:40am ET — HOWARD: "truly as fast as possible?" → honest NO; saturation corrected
  3. ~7:45am ET — HOWARD: "pretty fragile that a crash loses a day" → durability sweep run NOW
  4. ~7:50am ET — HOWARD DIRECTIVE: independent architecture review (drift concern)
  5. ~8:25am ET — TESTING WEDGE KILLED + BOTH ARCHITECTURE REVIEWS LANDED
  6. ~8:40am ET — HOWARD: reviews too technical; wants a paste-prompt for HIS OWN new thread
  7. ~9:00am ET — HOWARD (escalating): the REAL failure is the four inabilities; "don't waste my time"
  8. ~9:10am ET — HOWARD Q: if I grant the App.tsx exception, what stops the old free-for-all?
  9. ~9:15am ET — ⛔ HOWARD DECISION: App.tsx EXCEPTION GRANTED (single-use permit model)
  10. ~9:35am ET — permit-aware lint + both jobs relaunched; stale-fixture find
  11. ~9:50am ET — HOWARD (existential): "Is reliable AI development even realistic here, or do I hire people / shut it off?"
  12. ~10:00am ET — SAFE SHUTDOWN for Howard's restart
  13. ~10:05am ET — ⛔ CAUGHT MY OWN GPS BUG during shutdown: machinery branch NEVER pushed
  14. ~10:15am ET — FABLE VIABILITY REVIEW LANDED (pre-reboot) — guards attacked, four real holes
  15. ~10:35am ET — REBOOT RECOVERY COMPLETE (runbook executed)
  16. ~10:45am ET — HOWARD: "aghast at a week"; "start from scratch?"; "guards fail 40%?"
  17. ~10:55am ET — HOWARD asked for the full plan: priorities, parallel-vs-serial, who's on what
  18. ~11:20am ET — WAKE: GPT VIABILITY VERDICT LANDED + owner page LIVE + adapter patch drafted + stamper healed
  19. ~11:45am ET — ⛔ HOWARD RULING ×3 (escalating): NEW CODE ONLY, ENTRY-LEVEL BYPASS
  20. ~11:55am ET — WALLS BUILT + LIVE-TESTED; 7 BUILDERS RELAUNCHED; first App-free build compiling
  21. ~12:05pm ET — HOWARD commissions forward-only adversarial review (GPT-5.6)
  22. ~12:30pm ET — APP-FREE BUILD STAMPED + VERIFIED; walls hardened at the stamper; testing restaging
  23. ~12:45pm ET — FORWARD-ONLY VERDICT LANDED (relayed to Howard raw)
  24. ~1:00pm ET — HOWARD: sims all talking → FAB-93 silence rule ported to new shell

2026-07-22 from docs/reset/For Other Codex/SESSION_BOOK_20260722.md

  1. ~10:20am ET — FIRST VERIFIED MODULE-LIVE FLEET BUILD STAMPED
  2. ~10:40am ET wake — c29 torn-timing (benign), c30 = first real module cycle; panel published
  3. ~11:15am ET — HOWARD'S THIRD MEMORY CONFIRMED: the new router took over and was left half-done
  4. ~11:30am ET — HOWARD: audits missed the live router; amnesia = root cause; human-vs-LLM
  5. ~11:45am ET wake — module baseline BLOCKED by native-shell mismatch; true-layer fix building
  6. ~12:30pm ET — native-shell toolchain saga RESOLVED to a working recipe
  7. ~1pm ET — HOWARD ASKS re camera/barcode/label code → 4th abandonment case confirmed
  8. ~1:15pm ET — HOWARD: "keep searching, wire the best" → FULL-DISK ARCHAEOLOGY
  9. ~3:15pm ET — sweep DELIVERED (11 high-value rescues); shell chain: llama.xcframework restored
  10. ~4:20pm ET — NATIVE SHELL BUILT (chain10: SKIP_BUNDLING) + homed + stamper re-pointed
  11. ~4:30pm ET — FINAL ROOT CAUSE: .env never existed post-reboot → supabaseUrl undefined crash
  12. ~5:45pm ET — APP BOOTS (module-live) — root cause of canary stall = Speech Recognition dialog
  13. ~5:50pm ET — speech-grant timing handed to focused Composer (stop grinding synchronously)
  14. ~6:20pm ET — THE GATE OPENS: speech fix landed, fleet producing module-live traces
  15. ~6:40pm ET — HOWARD: "get it done by morning?" → OVERNIGHT BLITZ LAUNCHED
  16. HOWARD, before the overnight run: "Remember, this was already done before."
  17. ~10:30pm ET — BLITZ midpoint: fleet cycling clean, long pole committed, proofs = my queue

3. Direction changes

Any decision that reverses or reinterprets a Howard instruction must be written here the same day it is made.

DateWhat changedWhich instruction it affects Who decidedTold Howard?
2026-07-21 Fleet builds force food module LIVE; App.tsx frozen at dispatch new-code-first (restores it) conductor, executing Howard's standing order this ledger + chat, same day
2026-07-22 App.tsx declared RADIOACTIVE — delete-only, no line-count ratchet, reading-as-reference allowed; enforced at spec-lint + git pre-commit hook + register probe apptsx-frozen (strengthened) HOWARD this ledger + chat, same wake

Source: docs/reset/For Other Codex/DIRECTION_CHANGES.md · showing 2 of 2 row(s)

4. Legacy-file exception permits

The old main app file is frozen. A one-time exception can only land with a written permit. Status words: granted (approved, not locked yet) · armed (exact change locked, ready once) · consumed (already used) · expired (timed out).

P1-spine-adapter Granted

Howard approved a one-time exception, but the exact change is not locked in yet.

Granted by: Howard, chat 2026-07-23 ~9:15am ET (AskUserQuestion: 'Yes — grant it')

Scope: REVISED by Howard 2026-07-23 ~11:35am ET (supersedes shadow design): single DEFLECTION change — utterances are handed to the new command code and the old ladder does NOT process them; one env-driven setting (same pattern as the food-module flag) whose off-position restores legacy behavior as an emergency revert only; no side-by-side comparison mode. Howard verbatim: 'the one that deflects the utterances so the old code doesn't process them, and the new code is what processes them. New code. That is what we should test. If it's broken, we'll fix it.'

Expires: 2026-07-26T13:15:00Z

Used in commit: not yet

Internal status field: awaiting_diff

Source: scripts/lane-h/APPTSX_PERMITS.json

5. What changed since yesterday

One line per commit since 2026-07-22 (Eastern midnight), newest first.

  1. 7f0ea3f15 book: FAB-93 silence ported to new shell (1130483b7); restamp running
  2. 1beba698b book: forward-only verdict relayed; next-actions queue set
  3. 16186f67f stamper refuses legacy-shell bundles (verifyLegacyAbsent, 24/24); App-free build c1d8ddc36 stamped and verified; scheduler recovered
  4. 7ae699c03 book: forward-only adversarial review launched per Howard mandate
  5. 4093b1913 forward-only claim doc for Howard's adversarial review (decision + walls + known holes + his 4 questions)
  6. 5caff6760 entry walls: index.js legacy import removed upstream; entry files frozen (hook live-tested); 7 seam builders relaunched from c1d8ddc36
  7. a334adf21 HOWARD RULING: new code only, entry-level bypass — new shell forced in every build, App.tsx never mounts; old-path testing stopped; permit parked
  8. e263aaa4f wake: GPT verdict + owner page live + adapter patch under review + stamper healed
  9. 332f47344 adapter patch + worklog secured durably (787 lines vs base 35679d4d5); patch reviews dispatched
  10. 34a7ae973 book: full plan (serial spine + parallel tracks + roster) delivered to Howard
  11. df23bb4db book: Howard's aghast/scratch/40% questions + corrections
  12. 6306f8e72 reboot recovery complete: workshop rebuilt, snapshot branch on GitHub (fetch-back verified), jobs relaunched
  13. 53fbb4365 Fable viability review landed: hybrid ~60%, four guard holes found by live attack, phone-in-hand falsifier 07-30
  14. 1be43d308 book: masked-push GPS bug caught — machinery branch never on GitHub; scrub queued post-boot
  15. dacdddbe7 reboot runbook + safe-shutdown harvest (all worktrees pushed; adapter WIP patch saved) + book
  16. 064da7005 viability review mandate (Howard verbatim): live-fire guards, decision-vs-execution record, proceed/hire/hybrid/stop
  17. 96e6cb22c spec-lint permit-aware overrides (bare HOWARD-OVERRIDE hole closed) + stale conforming fixture fixed (17/17) + jobs relaunched
  18. da0adff7c permit machinery live (hook hash-gate + APPTSX_PERMITS P1 awaiting_diff; refusal demoed) + owner-page & adapter-draft jobs dispatched
  19. b4c08f43e HOWARD DECISION 2026-07-23: App.tsx single-use permit exception GRANTED (spine adapter; hash-pinned; demo-refusal-first)
  20. 1450f6a83 book: Howard permit-mechanism Q&A (exception = hash-matched single-use permit)
  21. ecef7b55d THE_FOUR_FAILURES doc (Howard's four inabilities: cause/guard/gap each) + review bundle
  22. 1fa8fc80c orchestrator position doc (plain language) + book: Howard's independent-assessor prompt request
  23. 66134162c testing wedge killed (sim-herd + canary host seesaw; stage-lane telemetry-first 36/36) + both independent architecture reviews landed
  24. 7e83e4f53 book: Howard architecture-review directive + claim doc + GPT-5.6 reviewer fired
  25. a2d689f9c ARCHITECTURE_TARGET_AND_PATH: the claim doc for Howard's independent reviews (drift register + 4 reviewer questions)
  26. 9ba53bd23 book: crash-durability sweep — all worktrees pushed, env backup verified
  27. 33bc5662e book: saturation corrected — 9 parallel jobs; mini needs address
  28. 0b1334701 book 07-23: telemetry-credentialed build live; measurement restored
  29. 7224e4ce7 book: blitz midpoint
  30. e82d26c48 book: Howard's header over the night — done is a tense, prove the present
  31. 223de06fa book: overnight blitz launched (pipeline2-exec long pole + F1 re-proof + 7 seam specs)
  32. 46e640f10 book: gate open — fleet producing module-live traces
  33. 720d1a8e7 stage-lane: effective Speech Recognition grant (TCC write + tccd reload)
  34. 689d9366f book: speech-grant timing → focused Composer; milestone (app boots) stands
  35. 0a1ca26ba book: app boots module-live; speech-recognition TCC fix; toolchain recipe
  36. 432e4a3d1 stage-lane: grant Speech Recognition via TCC.db (simctl privacy lacks the service); dead-pid locks stale immediately
  37. c151bfab7 isProcessAlive: EPERM means alive (latent bug exposed by dead-pid-stale fix)
  38. 3bd7ced6f locks: dead-pid locks are stale immediately (shared isLockStale; ends manual lock deletion after daemon kills)
  39. e82970391 book: supabaseUrl root cause + full env injection; env-complete build in flight
  40. 163166c43 stamper: inject ALL EXPO_PUBLIC_* from worktree .env at compile (supabaseUrl-undefined crash root cause)
  41. 0988adaff book: native shell built+homed; stamper re-pointed; awaiting stamp
  42. b6d3bdde2 stamper: shell app at reboot-immune /Users home (fresh xcodebuild shell, llama embedded)
  43. 530530b35 book: sweep rescues (11 HIGH); llama framework restored; chain8 running
  44. 1c7ce3f01 book: full-disk archaeology launched (~70 dirs); pass-3 report landed
  45. 9a59748fd book: photo/scan modules = 4th unmerged abandonment; RESC1 queued
  46. 1427ef60f book: native toolchain recipe (ruby@3.2 + isolated gems) — shell build running
  47. 31b9ff9bb book: llama crash = old-shell/new-JS mismatch; fresh native shell building
  48. 82bbd9c47 book: pass-3 runtime cross-exam launched; ARCH1 living-ledger queued
  49. 858a293fe book: router live_all never reverted — execution half never finished; priority change
  50. fd881dab9 book: c30 first module cycle; panel published; F1 held for re-proof
  51. d123db2d4 GROUND TRUTH panel: which-code-ran %, instruction register, build provenance (Codex build, conductor-committed after tip-check)
  52. d2c1227ea FIRST MODULE-LIVE BUILD STAMPED; register probe artifact-level GREEN
  53. e4e77aeda book: inlining solved, shell restored, decisive module-live build in flight
  54. 1b3dcaced test: bad fixture = gate-off form
  55. d291251f0 stamper: GATE_ON_MARKER matches real compiled form (t="1") — proven by manual probe
  56. 1a8c57cb5 book: inlining mechanism probe running; fleet stalled-by-design behind the gate
  57. a59d28119 book: capability maps delivered; cutover-F1 + Howard panel launched
  58. 96549bf30 capability maps: working-tree + git archaeology (Howard asset inventory)
  59. 47ab6b292 book: radioactive enforced; asset inventory launched per Howard
  60. e659aaed1 HOWARD RULING: App.tsx radioactive — delete-only (lint + hook + probe + ledger)
  61. a5a66131d book: reboot recovery — worktree rebuilt, launchd installed, module-live build pending
  62. 347bcfe68 stamper: --reset-cache on export:embed (stale-transform guard for env inlining)