Libretto
Roadmap
Feature inventory · 160 items · 13 tracks
-
160
Total rows
153 still in the plan · 7 dropped
-
36
Verified on device
17 built, not device-verified
-
0
In progress
Started, not judgeable yet
-
100 / 0
Not started / blocked
5 XL rows still unsplit
-
57
Need Dylan
No session can close these alone
- 36 verified on device (23.5%)
- 17 built, awaiting the device (11.1%)
- 100 remaining
- 7 dropped, out of the denominator
Where we are
The next session of each run, the one row a session could open the repo and start today, and everything wrong with the build Dylan is already holding.
H1 — soundtrack lab
Session 5 next · 3 rows · 21 sessions in the run
ST14 · SC14
Desk prototype
Session 300 next · 2 rows · 12 sessions in the run
D11 · D24
PCB + enclosure
Session 400 next · 3 rows · 6 sessions in the run
D15 · D17 · D20
Shelved — public app, deferred
Session 102 next · 1 rows · 9 sessions in the run
PL9
- Next actionable
- G2 · Gaze accuracy log + go/no-go note for the device concept (S · S7)
- The floor
- Nothing is wrong with the APK Dylan holds.
How to read this
States
- Verified on device 36
- Built, and a human confirmed it on the physical device.
- Built, not device-verified 17
- In the repo, tests green, only the emulator has seen it. Looks done, is not.
- In progress 0
- Started, incomplete, not judgeable yet.
- Planned 100
- Has a cut point, not started.
- Blocked 0
- Cannot move until something outside the row moves.
- Dropped 7
- Decided against. Kept so nobody re-files it; never counts as progress.
Effort
- S 43 open
- fits alongside other work
- M 48 open
- most of a session
- L 21 open
- a full focused session
- XL 5 open
- does not fit in one session — must be split before it is numbered
5 XL rows are still unsplit — a planning defect, not work: D4, D5, D6, D7, D8.
Flags
- Floor
- Wrong or missing on the APK Dylan already holds. Not the bar — the floor.
- Needs Dylan
- A decision, an account, a purchase or a sitting. No session can finish it alone.
- Ultra review
- Warrants /code-review ultra. A recommendation, never auto-run.
- Ultracode
- Run the session with the ultracode multi-agent workflow.
Who runs the session
- Fable
- Planning sessions only: builds the plan, makes sure Opus follows it.
- Opus
- Opus takes charge of the whole session.
- Opus + Sonnet
- Sonnet builds the mechanical parts, Opus supervises and reviews the diff.
Roadmap, dashboard & process
Verified 90.9% · 11 rows
This site, the session loop (/start, /end), hygiene and research.
- 10 verified on device
- 1 built, not device-verified
- M1 Blueprint: VISION, ARCHITECTURE, SCORE_TIMELINE_SCHEMA, PREPROCESSOR, ROADMAP L
Cut point: Docs committed and cross-linked; stack pinned (Flutter, Android first).
More
Evidence:
f0a390d, 72209daProved by: docs · 2026-06-21
- M2 Sideloadable demo APK (release, debug-signed; no ADB needed) S
Cut point: Dylan installs it on the S25 without a computer.
More
Evidence:
0704cc3 · build/app/outputs/flutter-apk/app-release.apk (v4 on 2026-07-05)Proved by: installed; no clear verdict yet (2026-09-20) · 2026-07-04
- M3 Outside product review folded into the docs (6 points) S
Cut point: Sustained-read exit, loop-fatigue sourcing, Phase-4 gate, H2 hypothesis, H2.5 rung, LFS/OGG flags all in the docs.
More
Evidence:
0704cc3Proved by: docs · 2026-07-04
- M4 Jev (TypeSafe AI) research + the two-mode decision M Ultracode Fable high S0
Cut point: Report saved under docs/research; two-mode rule in ARCHITECTURE; JevTagger row in the plan.
More
Evidence:
docs/research/JEV-2026-09-20.mdProved by: docs · 2026-09-20
- M5 Strategy research: soundtrack generators, arrangement, artists, competition, hardware, licensing L Ultracode Fable high S0
Cut point: Brief saved with verified/refuted claims; decisions written into the plan rows below.
More
Evidence:
docs/research/STRATEGY-2026-09-20.mdProved by: docs · 2026-09-20
- M6 Roadmap site (this) — inventory, derivation, brief, session plan, paper theme L Ultracode Fable high S0
Cut point: `npm run gate` green; both views render; every session card carries MODEL/EFFORT/ULTRACODE/ULTRA-REVIEW tags and a cut point.
More
Evidence:
roadmap/ · npm run gate green (23 tests, astro check 0/0/0) · release 20260920192748Proved by: gate + https://libretto.tinkr.es/sessions 200 from outside · 2026-09-20
- M7 Management dashboard at /admin (books & licences, stems, competition, research) M Fable high S0
Cut point: Four data-driven admin pages behind nginx basic auth.
More
Evidence:
roadmap/src/pages/admin/* · https://libretto.tinkr.es/admin/Proved by: /admin/ = 401 without password, 200 with; four pages render · 2026-09-20
- M8 libretto.tinkr.es live: Cloudflare A record, nginx block, certbot, deploy script with neighbour check M Fable high S0
Cut point: `curl -sI https://libretto.tinkr.es` = 200 from outside; /admin/ = 401 without password; neighbours unchanged.
More
Evidence:
deploy/deploy-roadmap.sh --cert · cert expires 2026-12-19Proved by: curl from outside: / 200, /admin/ 401; neighbours 200/302/200/200 · 2026-09-20
- M9 /start and /end wired to the plan (digest script, brief-first briefing, inventory step, Sonnet cleanup agent) M Fable high S0
Cut point: `bash .claude/scripts/start-check.sh` prints "Session 1 next" in one call; /end moves rows + regenerates + redeploys.
More
Evidence:
.claude/scripts/start-check.sh · ~/.claude-personal/commands/{start,end}.md · .claude/agents/cleanup.mdProved by: digest prints the brief first in one call · 2026-09-20
- M10 Context hygiene: CLAUDE.md ≤ 6 KB, GOTCHAS.md, one-line memory index, ROADMAP.md as history S Fable high S0
Cut point: Before/after byte counts in BUILD_LOG; nothing lost (grep-able).
More
Evidence:
BUILD_LOG 2026-09-20: CLAUDE.md 15,943→4,367 B; MEMORY.md 11,434→1,964 B; docs/GOTCHAS.md 14,118 BProved by: byte counts + grep of moved titles · 2026-09-20
- M11 Standing rule: every question through the AskUserQuestion tool S Fable high S0
Cut point: One line in the global CLAUDE.md + a feedback memory.
More
Evidence:
memory/feedback-questions-via-question-tool.mdProved by: docs · 2026-09-20
Sonic core (Phase 1)
Verified 42.1% · 22 rows
Real beds in the mixer and the on-device verdict that gates everything downstream.
- 8 verified on device
- 5 built, not device-verified
- 6 planned
- 3 dropped
- SN1 flutter_soloud mixer: all voices alive, 16 ms equal-power crossfade ticker L
Cut point: A crossfade between two looping stems feels like a score transition on a device.
More
Evidence:
ffbae74 · lib/mixer/audio_mixer.dartProved by: Pixel 7: Dylan — "the transition feels good" · 2026-06-22
- SN2 Cue sequencer with 1200 ms hysteresis (no mix flapping) S
Cut point: Jitter test passes; no audible flapping on device.
More
Evidence:
ffbae74 · lib/mixer/cue_sequencer.dartProved by: jitter test + Pixel 7 · 2026-06-22
- SN3 Mixer debug screen with live per-stem gain bars + source-family selector S
Cut point: Gains visible live; family switch rebuilds the mixer.
More
Evidence:
ffbae74, 3358344 · lib/app/mixer_debug_screen.dartProved by: Pixel 7 · 2026-06-22
- SN4 Placeholder stem generator (seam-free synthetic WAV pads) S
Cut point: 7 loop-safe placeholder beds render and loop without clicks.
More
Ugly on purpose; re-voiced caricature-distinct on 2026-07-05 so timing can be judged.
Evidence:
ffbae74, ae225c5 · tool/gen_placeholder_stems.pyProved by: Pixel 7 + emulator · 2026-06-22
- SN5 stems.json v2 + source-family A/B harness (placeholder / cc0 / ai, role ids family-agnostic) M
Cut point: Swapping a family needs zero change in cue/sequencer/mixer code; families A/B live from the debug screen.
More
Evidence:
3358344 · lib/stems/stem_bank.dart, assets/stems/stems.jsonProved by: analyze/test/APK green; never heard with real audio · 2026-06-22
- SN6 Seamless-loop prep tool (equal-power tail→head crossfade) S
Cut point: Any raw clip becomes a click-free loop.
More
Evidence:
3358344 · tool/make_seamless_loop.pyProved by: synthetic input only · 2026-06-22
- SN7 Sourcing pack: Freesound CC0 searches + ElevenLabs SFX prompts, D-minor beds S
Cut point: Dylan can source a full role set from the document alone.
More
Evidence:
3358344, 0704cc3 · tool/phase1_real_stems_sourcing.mdProved by: doc · 2026-06-22
- SN8 80 scene-level cues for Frankenstein + caricature-distinct placeholder voices M
Cut point: Mood changes land mid-chapter at story beats; moods sound different from each other.
More
Evidence:
ae225c5 · tool/gen_frankenstein_timeline.pyProved by: emulator: cue flipped mid-chapter; S25 verdict pending · 2026-07-05
- SN9 make_seamless_loop.py: downbeat-snap pre-pass (allin1) + concat mode for multi-take beds M Opus + Sonnet high S1
Cut point: 2-3 short takes stitch into one 60-120 s bed with loop points on downbeats; tests on synthetic takes.
More
allin1 + Demucs are MIT, and allin1 stays OPTIONAL (no Python 3.14 wheels) - the numpy spectral-flux fallback is what every test runs on. Never madmom pretrained weights (CC-BY-NC-SA). First real audio through it is ST1.
Evidence:
tool/loopkit.py, tool/make_seamless_loop.py, tool/tests/ (42 tests)Proved by: 3 misaligned 24 s takes -> a 97.81 s bed, 48.00 bars, wrap +0 ms off the downbeat, seam ratio 0.33 · 2026-09-20
- SN10 APK v5 with the four AI families S Needs Dylan Opus high S2
Cut point: Release APK built and handed to Dylan.
More
Reader now follows stems.json defaultFamily (= ai_sa30), so reading plays real beds rather than the synthetic pads. 53 MB of the 92.7 is three ABIs of native libs - PL10 (per-ABI split, S11) would roughly halve it.
Evidence:
build/app/outputs/flutter-apk/app-release.apkProved by: 92.7 MB; rebuilt from HEAD and installed on the S25 over wireless adb 2026-09-23, all four families played layered in the mixer and the reader · 2026-09-23
- SN19 Pop-free crossfades: sine-law equal-power entry + audio-thread volume ramps S Opus high S2
Cut point: Dylan steps cues on the S25 and hears no pop or lurch at a cue change.
More
Unplanned, found in the S2 listen. The old curve sqrt((1-p)a² + p·b²) has an infinite slope at p=0: a bed entering a 3 s fade leapt from silence to about -23 dB on its first 16 ms tick. And setVolume stepped the gain per Dart tick, so it zipped and lurched whenever the UI isolate stalled. Now a·cos + b·sin (same constant power, finite onset) handed to SoLoud fadeVolume over 24 ms. Switching FAMILY on the debug screen still clicks: that is a full engine restart and exists only there.
Evidence:
lib/mixer/audio_mixer.dart crossfadeGain, test/crossfade_gain_test.dartProved by: Dylan on the S25 2026-09-23: "gone"; 44 Dart tests green · 2026-09-23
- SN20 Streaming mixer: only the current and next cue beds are loaded, from disk L Needs Dylan Opus high S4
Cut point: The mixer loads/streams just the beds the current and the upcoming cue need (not every bed of the family); a 32-bed pack keeps native heap near today (~150 MB on the S25); the loop seam and pop-free crossfades are re-proven by ear on the S25.
More
Today init() decodes EVERY bed into RAM and plays them all at volume 0: 148 MB native heap measured for 8 beds of 89 s on the S25. 32 beds of 3-4 min would be ~1.1-1.3 GB, and the e-ink target has 2-4 GB total. A bed that re-enters should start at a random offset, since silent background playback (which gives that today) goes away.
Evidence:
lib/mixer/audio_mixer.dart, lib/mixer/mixer_engine.dart SoLoudEngine (LoadMode.disk), ScoreConductor prefetch hint, test/audio_mixer_test.dart, tool/make_stress_pack.pyProved by: S25 2026-09-24: native heap peak 38 MB reading DCC 15 min (148 MB in S3); 36 MB under a 32-bed x 4.5 min stress pack (56 MB) paged fast, 35 bed entries, 0 open failures; seeks 5-30 ms. Dylan by ear: loops, crossfades and fast page turns clean, no silence or cuts. · 2026-09-24
- SN21 Streaming mixer re-measured on the e-ink target: bed-entry seeks and loop wraps under the audio lock S Needs Dylan Opus high
Cut point: On the D11 e-ink device, libretto.mixer seek timings and a short-loop stress pack (8 s beds, wraps every 8 s) are logged; either every seek stays well inside the audio buffer and nothing clicks by ear, or the fix (copy beds to internal storage, pre-seek, or memory mode for the few live beds) is chosen from the numbers.
Waiting on:
D11· Real APK sideloaded on a borrowed AOSP e-ink device (OS-answer test)More
SoLoud seek() and a disk stream's loop-wrap seek both hold the mix thread's lock and do file I/O. S25: 5-30 ms per entry seek (internal storage 3-17 ms, so mostly decode), inside the ~43 ms buffer and clean by ear. A slower e-ink SoC could underrun. make_stress_pack.py --family placeholder_ogg --min-seconds 1 builds the short-loop variant.
- SN22 Moment cues: one-shot sounds fired on their sentence (a blast, a door shaking) M Needs Dylan Opus high S6
Cut point: The timeline carries moments (anchor char offset + one-shot file + gain); the conductor fires each once when its anchor comes into view; the mixer plays one-shots over the beds; unit-tested and heard on the S25 on at least 3 DCC moments.
Waiting on:
SC14· DCC v2 timeline: place + feel per beat, about twice the beats, the place layer steadyMore
Dylan, 2026-09-24: a one-off event must not live inside a loop that repeats it every few minutes. Position is page-level today, so a moment lands on its page, not its word, until gaze.
Evidence:
lib/timeline/score_timeline.dart (Moment), lib/timeline/score_conductor.dart (fireMomentsIfDwelt), lib/mixer/audio_mixer.dart (playOneShot), lib/timeline/score_pack.dart (oneShots), tool/packkit.py, tool/gen_book_timeline.py --momentsProved by: flutter analyze clean, 107 Dart + 158 Python tests; not yet heard: no DCC moments are tagged and no one-shots are rendered · 2026-09-24
- SN11 CC0 family: Dylan's Freesound picks per role, loop-prepped, provenance in stems.json M
Cut point: 7 CC0 beds with Freesound ids + licence recorded.
More
DROPPED 2026-09-24 (S6 triage, per-book pivot): A CC0 family for the generic bank. Scores are per book (2026-09-23) and every DCC bed comes from Eleven Music v2 or Stable Audio 3.0.
- SN12 Source-family A/B verdict on the S25 (CC0 vs best AI family) M
Cut point: A family is chosen and the reason written to memory + BUILD_LOG.
More
DROPPED 2026-09-24 (S6 triage, per-book pivot): The source A/B it was for is decided: Eleven Music v2 + Stable Audio 3.0, per book (ST1 ear verdict, S3-S5).
- SN17 OGG q4 vs WAV listening A/B on the S25 (placeholder vs placeholder_ogg) S Needs Dylan Opus high S6
Cut point: Dylan cannot pick the OGG family out on the S25 - or q6 is adopted and the budget in docs/AUDIO_ASSETS.md is updated.
More
Split out of PL6 (S1): the format decision is made and evidenced; only the ears call is left, and it belongs with the other S4 device verdicts on real beds.
- SN13 Mood-at-right-beats verdict, round 2 (80-cue timeline on real beds) S
Cut point: Dylan confirms changes land at story beats and moods are distinct, or a named fix list exists.
More
DROPPED 2026-09-24 (S6 triage, per-book pivot): The Frankenstein 80-cue round. Superseded by the DCC blind feel-fit bench (S5) and SC15, the v2.1 read on the S25.
- SN14 Phase 1 exit: 45-60 min sustained read without loop fatigue M Needs Dylan Opus high S6
Cut point: Dylan reads 45-60 min on the S25 and the score never pulls him out; Phase 1 declared closed.
Waiting on:
SC15· DCC v2.1 heard on the S25: place-first, quiet, not repetitive, and fitting the passageMore
THE gate: Phase 4 (SC4+) does not start before this.
- SN15 APK v6 (reader polish + chosen family as default) S Opus + Sonnet high S8
Cut point: Release APK built and handed to Dylan.
- SN16 APK v7 (library + polish) S Opus + Sonnet high S13
Cut point: Release APK built and handed to Dylan and the test group.
- SN18 Stinger asset family + one-shot mixer layer M Opus high S19
Cut point: A stinger asset family exists (source + license recorded like any stem) and the mixer plays a one-shot stinger over the current bed with no crossfade, without disturbing the bed's own loop.
Waiting on:
P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it again
Soundtrack program
Verified 50% · 16 rows
AI generators, adaptive arrangement, commissioned packs, author-recommended artists.
- 7 verified on device
- 1 built, not device-verified
- 6 planned
- 2 dropped
- ST1 Generator shoot-out: Stable Audio 3.0, Stable Audio 3 Small (local), Eleven Music v2, ACE-Step — 7 roles each L Needs Dylan Opus high S2
Cut point: Families ai_sa30 / ai_sa3 / ai_11m / ai_ace in stems.json, loop-prepped, selectable from the debug screen.
More
Stable Audio 2.5 was dropped for Stable Audio 3.0: the web app Dylan subscribes to now runs 3.0, and 2.5 is reachable only through a separate paid platform API - testing a superseded model for 10 USD more. ACE-Step is DISQUALIFIED on the numbers (0/7 beds layer without a clashing third) despite being the only Apache-2.0 entrant. The ear verdict (2026-09-23) splits from the key measurement: ai_11m beat ai_sa30 layered, so a clean third is necessary (ACE fails it audibly) but not what ranks the survivors.
Evidence:
tool/generate_beds.py, tool/genkit/, assets/stems/ai_*Proved by: 28 beds, every seam +0 frame drift; flutter analyze clean, 41 Dart + 109 Python tests. HEARD on the S25 2026-09-23, layered: Dylan picked Eleven Music v2 as best layered (it measured 0/4 in D minor); only ACE-Step clashed audibly · 2026-09-23
- ST2 Key-adherence + cost report per generator (D minor held? USD per bed, minutes per bed) S Opus high S2
Cut point: docs/research/GENERATORS-<date>.md with a table; spend ≤ 30 USD.
More
Answers the open question from the S0 strategy sweep: generators DO differ on key. Stable Audio 3.0 held D minor on 4/4 tonal roles, Eleven Music on 0/4, ACE-Step on 0/4 with every bed clashing. Prompting harder does not fix it (ai_11m_kx).
Evidence:
docs/research/GENERATORS-2026-09-20.md, generator-ledger.jsonl, generator-keyfit.jsonProved by: Generated from the ledger, not written by hand. Spent 0.20 USD real / 2.90 committed of 30. · 2026-09-20
- ST7 Best-of-N take selection: generate 3 takes per role, keep the one keyfit says is in key S
Cut point: generate_beds.py --takes 3 --pick-best chooses by measured third; a re-run of ai_11m raises its layers-without-clashing score.
More
DROPPED 2026-09-24 (S6 triage, per-book pivot): The selection mechanism shipped as ST12 (--max-tries, used for every DCC bed); what was left is re-running the generic ai_11m bank, which only matters for a book with no pack. Before: The actionable half of the ST2 finding. Restating the key in the prompt does nothing (measured), but the harness already measures every bed - so generate a few and keep the one that landed. Cheap on a subscription, and it turns key adherence from a hope into a filter. S3: the selection mechanism shipped as ST12 (--max-tries); what is left is this cut, the ai_11m re-run, which matters only for the generic fallback bank now that scores are per book.
- ST8 Blind A/B listening bench for stem families S Opus high S2
Cut point: Families compared blind, sides shuffled per role, verdict recorded as evidence beside the measurements.
More
Unplanned, built on Dylan’s request to be part of the measurement. keyfit cannot say whether a bed sounds like a score, so the ear is evidence too - collected blind because knowing which one cost money decides a listening test without anyone noticing.
Evidence:
tool/make_ab_page.py, docs/research/generator-verdicts.jsonProved by: First round recorded 2026-09-20: 3 ties, 2 ai_sa30, 2 ai_11m over 7 roles · 2026-09-20
- ST9 Textural beds that actually sit on D (drones and pulses lose the root) M
Cut point: At least 2 of the 3 textural roles measure a D root above 15% chroma share in the chosen family.
More
DROPPED 2026-09-24 (S6 triage, per-book pivot): Generic-bank textures. The per-book place beds are now pure field-recording ambience with no key at all (v2.1), so there is no root to hold. Before: Every generator did worst here: best score was 1/3 textural roles rooted on D. dread_drone in particular came back at 0-1% root in three of the four families. The tonal roles are solved; the textures are not.
- ST10 Per-book score design: Dungeon Crawler Carl (world, characters, emotions → stem roles, key, cue vocabulary) M Needs Dylan Opus high S3
Cut point: docs/research/SCORE-DCC.md names the stem roles (places, characters, System AI, combat), the key/tonal plan, the prompts, and where the book-derived files live on the device; Dylan approves it.
More
The roles come from the book, not from a generic mood list. The design doc may name the book; it must not quote it at length (public repo).
Evidence:
docs/research/SCORE-DCC.mdProved by: Dylan approved the roles, key plan and mood vocabulary 2026-09-23, before any spend · 2026-09-23
- ST11 DCC stem family on Eleven Music v2: generated, loop-prepped, seam- and third-checked M Needs Dylan Opus high S3
Cut point: A book family (e.g. dcc_11m) exists via generate_beds.py with every seam +0 frames and no bed carrying a clashing third; spend stays inside the 30 USD cap.
More
Generator chosen by ear 2026-09-23 (ai_11m best layered). Roles for a book are no longer the fixed seven, so generate_beds.py/roles.py must take a per-book role set.
Evidence:
assets/stems/dcc_11m/ (8 beds), generator-keyfit.json dcc_11m, ledger rows dcc_11mProved by: 8 beds from 11 takes, every seam +0 frames, no kept bed carries F-sharp; 2.48 USD list (ledger 5.38 of the cap); Dylan on the S25: nothing grated · 2026-09-23
- ST12 Re-roll on a clashing third: generate_beds.py --max-tries keeps the first bed that layers S Opus high S3
Cut point: --max-tries N re-rolls a bed whose third reads major or both, keeps the first that layers (else the least F-sharp), and a real family ships through it.
More
Split from ST7 in S3, because the DCC cut needed it. A generator error is never retried, only a clashing third, so a flaky API cannot spend the cap in a loop. Blind spot found on the first real run: a bed that moves its whole tonal centre (donut_regal came back in A-flat) has neither F nor F-sharp and passes as absent; ST14 adds a tonal-centre gate.
Evidence:
tool/generate_beds.py make_bed/better, tool/tests/test_book_score.py ReRollsProved by: dcc_11m: tension_stalk, system_gloss and grief_earth re-rolled once each, all 8 kept beds layer; Dylan on the S25: nothing grated · 2026-09-23
- ST13 DCC score v2 design: an environment bed per place + a feel bed per emotion/character M Needs Dylan Opus high S5
Cut point: SCORE-DCC.md v2 lists every distinct place in book 1 (counted from the text, ~15-25) and ~12 feel beds (the 8 moods + awe, triumph, character motifs) with their prompts and lengths (env 3 min, feel 4 min); Dylan approves it.
More
Dylan on the S25, 2026-09-23: the 8-bed score is too repetitive; one bed per environment, feel and emotion, way more than 8, and 4-min loops. Two layers make every place x feel combination distinct with ~32 beds instead of one bed per combination.
Evidence:
docs/research/SCORE-DCC.md "v2 design" · tool/genkit/roles.py DCC2_FEEL/DCC2_ENVProved by: 476 beats from a full two-axis re-read by four Opus readers; 38 place slugs merged to 20 place beds + 15 feel beds; Dylan approved the design and spend before any call (2026-09-24) · 2026-09-24
- ST14 DCC v2 bed library on Eleven Music: ~32 beds, third- and tonal-centre-gated M Needs Dylan Opus high S5
Cut point: Every ST13 bed exists via generate_beds.py (--max-tries), seams +0 frames, no clashing third, tonal beds D-minor rank <= 6; spend inside the 60 USD cap (estimate ~18 USD with re-rolls).
More
Cap raised 30 -> 60 USD by Dylan on 2026-09-23. S5: the ElevenLabs credit QUOTA bound, not the USD cap (Starter 40k, then Creator 121k, both exhausted; ~858 credits/min measured). 28 beds Eleven, 7 Stable Audio 3.0 (web app, imported). Exceptions kept: feel_system rank 8, feel_triumph rank 9 (no F sharp). Beds were only peak-normalised (21 dB loudness spread) - now levelled to -20 dBFS RMS.
Evidence:
private/beds/dcc2_11m (35 beds) · generate_beds.py passes()/--resume/--keep · loopkit.levelProved by: 35/35 beds, seams +0; blind desk bench v2 7 wins vs v1 1 (4 neither), fit 2.92 vs 2.33; not yet heard on the S25 · 2026-09-24
- ST15 DCC v2.1 place beds: 20 pure field-recording ambiences on Stable Audio 3.0 S Needs Dylan Opus high S6
Cut point: All 20 DCC2_ENV beds re-made from the v2.1 prompts (no drone, no music, 3 min) on Stable Audio 3.0, imported with generate_beds.py --generator import, levelled, seams +0, published in dcc2_11m.
Waiting on:
ST14· DCC v2 bed library on Eleven Music: ~32 beds, third- and tonal-centre-gatedMore
Dylan, 2026-09-24 after the bench: quiet and place-first - the sound of the space leads, music so low you almost stop reading to hear it. Prompts are ready in audio-masters/dcc2_sa30/v21/PROMPTS.txt (most-used place first); Dylan renders them by hand in the web app (no API).
- ST16 DCC feel beds that fit: system, show, defiance, respite and guide re-briefed and re-made S Needs Dylan Opus high S6
Cut point: The five feels that scored 1-2 on the blind bench get new prompts and Stable Audio 3.0 takes; a re-run of the bench scores each at 3 or better.
Waiting on:
ST14· DCC v2 bed library on Eleven Music: ~32 beds, third- and tonal-centre-gatedMore
System and show scored 1 in BOTH scores, so it is the brief, not the key: the cheerful-sinister game layer has never landed. S6: the five briefs are rewritten in roles.py (system = indifferent hold music, show = a hollow house band, respite = exhausted relief, guide = gruff not sentimental, defiance = a punk riff, not an anthem); the Stable Audio renders wait on Dylan (audio-masters/dcc2_sa30/v21/PROMPTS-ST16.txt).
- ST3 ScoreArrangement design (horizontal re-sequencing on SoLoud 4.1 scheduled play + loop points) for artist tracks S Opus xhigh S10
Cut point: ARCHITECTURE section written: segments (allin1/Demucs) → per-mood loops + transitions; what stays hand-authored.
- ST4 Composer commission #1 (work-for-hire): 8-10 roles, 60-120 s evolving beds, 24-bit/48k WAV, key/tempo sheet, stingers L Needs Dylan Opus high S108
Cut point: Pack delivered and accepted; contract signed (WFH, derivative-work right explicit); 5-20k USD band.
Waiting on:
SN14· Phase 1 exit: 45-60 min sustained read without loop fatigueMore
Indie-tier rate anchor 400-2,000 USD per finished minute (GameSoundCon 2025).
- ST5 Arrangement of the commissioned pack through the ScoreArrangement pipeline M Opus high S108
Cut point: Pack becomes a stems.json family and drives a whole book.
Waiting on:
ST3· ScoreArrangement design (horizontal re-sequencing on SoLoud 4.1 scheduled play + loop points) for artist tracks - ST6 Author-recommended artist outreach (first author + first artist agreed in principle) M Needs Dylan
Cut point: One author names an artist; one artist says yes to a scoped pack.
Waiting on:
CA8· Artist letters: author-recommended artists shortlist + outreach template
Reading experience
Verified 31.3% · 16 rows
EPUB rendering, reading position, comfort, the reader UX (Phases 2-3 follow-through).
- 5 verified on device
- 3 built, not device-verified
- 8 planned
- R1 EPUB reader (flutter_epub_viewer / epub.js), WebView isolated to lib/reader L
Cut point: Seed EPUB renders and paginates on a device.
More
Evidence:
b29ed63 · lib/reader/reader_screen.dartProved by: Pixel 7 · 2026-06-23
- R2 ReadingPositionSource keystone: chapterId = spine-CFI prefix, percentThroughBook = epub.js progress M
Cut point: Position survives a font reflow; hermetic unit tests green.
More
Evidence:
b29ed63 · lib/position/Proved by: Pixel 7: 50.00% / /6/30 held across reflow; 11 tests · 2026-06-23
- R3 Home screen (Mixer | Reader) + mixer bootstrap moved verbatim S
Cut point: Both screens reachable on device.
More
Evidence:
b29ed63 · lib/app/home_screen.dartProved by: Pixel 7 · 2026-06-23
- R4 Chapter-anchored cue lookup (`bookFromChapter`) with fallback to book percent M
Cut point: Conductor prefers chapter geometry when chapter ids match; non-regressive otherwise.
More
DORMANT: reader emits CFI-spine ids (/6/30), timeline uses slugs (chapter-5). R7 lights it up.
Evidence:
12fe1a1Proved by: 30 tests · 2026-07-01
- R5 Open from device: file_picker → plain reader for personal EPUBs (no score) S
Cut point: A picked .epub renders; personal books never enter the repo.
More
allowScriptedContent:true is a live surface for user-supplied books → PL8 before H2.
Evidence:
12fe1a1Proved by: analyze + tests · 2026-07-01
- R6 ReaderTheme (bg + ink always together) + reading UX redesign (comfort sheet, score strip, debug toggle) M
Cut point: Text renders in Dark/Sepia/Light; theme switches live; debug chrome hidden by default.
More
Evidence:
3d98829 · lib/reader/reader_theme.dart, test/reader_theme_test.dartProved by: emulator release; regression test · 2026-07-05
- R14 Runtime score pack: an opened EPUB finds its own timeline + stem family M Needs Dylan Opus high S3
Cut point: Opening DCC from the device plays its own score on the S25; a book with no pack opens with the default family and no cues, never a crash.
More
Today the scored reader hard-codes frankenstein.epub + its timeline asset. Match on the EPUB identity (dc:identifier / hash), not the filename.
Evidence:
lib/timeline/score_pack.dart, lib/app/scored_reader_screen.dart ScoredBook, tool/push_score.py, test/score_pack_test.dartProved by: S25 2026-09-23: DCC opened from the device found its pack by SHA-256 and played its own score; a book with no pack opened in the plain reader · 2026-09-23
- R15 Score packs off the APK: a book pack carries its timeline AND its beds, pushed to the phone M Needs Dylan Opus high S4
Cut point: dcc_11m leaves assets/ and the APK; push_score.py pushes timeline + beds + a pack manifest to the app files dir; StemBank loads the family from disk and DCC plays on the S25 as it does today.
More
v2 is ~32 beds, ~55 MB: in the APK it breaks the 60 MB audio budget and the no-LFS rule. A per-book pack is personal anyway (it is keyed to one EPUB), so it travels the way the timeline already does.
Evidence:
tool/packkit.py, tool/push_score.py, lib/timeline/score_pack.dart, lib/stems/stem_bank.dart PackStems, test/score_pack_test.dart, tool/tests/test_packs.pyProved by: S25 2026-09-24: dcc_11m pack (8 beds + timeline + pack.json) pushed, swapped in and size-verified; DCC opened from the device played its score; Dylan: "clean". APK 102.8 -> 97.6 MB. · 2026-09-24
- R16 Companion mode: the score plays on the phone while Dylan reads on his Kindle M Needs Dylan
Cut point: With DCC open on a Kindle and Bluetooth headphones on the phone, the score advances at a calibrated reading pace and resyncs from a typed Kindle percent; drift stays inside one cue between check-ins.
More
Dylan, 2026-09-23: read the whole book on the Kindle (easier on the eyes), hear the score on headphones. A stock Kindle exposes no reading position (no API, cannot run the app), hence pace + check-ins; KOReader on a jailbroken Kindle could report page turns; an Android e-ink reader runs Libretto as-is (D11). Captured only: Dylan places it after v2.
- R7 Reader CFI-spine → timeline-slug mapping (lights up the chapter-anchored lookup) M Opus + Sonnet high S8
Cut point: Conductor takes the chapter path on the real Frankenstein EPUB; verified against its CFIs on device in the next sitting.
- R8 Font-size change debounce S Opus + Sonnet high S8
Cut point: Rapid font taps no longer race the async reflow.
- R9 Reading-speed smoother on the position source S Opus + Sonnet high S8
Cut point: A fast page-flip does not jump the mix.
- R10 Default ambient bed for device-opened (unscored) books S Opus + Sonnet high S8
Cut point: Opening a personal EPUB plays a calm bed instead of silence.
- R11 Small cleanups: `.single` → `.firstOrNull` in open-from-device; Phase-2 debug app bar off the scored screen S Opus + Sonnet high S8
Cut point: Both gone; analyze + tests green.
- R12 Renderer char-offset<->screen-rect map M Opus high S18
Cut point: For the current layout (font size, margins), the reader exposes a char-offset<->screen-rect map so a screen coordinate (tap or, later, gaze) resolves to a word and its char-offset; verified across at least two font sizes on device.
Waiting on:
P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it againMore
Amendment (Dylan, 2026-09-23): reflow already keys cues to char-offset, never pixel CFI; this is the missing half — screen coords back to char-offset — that word-level gaze (and tap-to-trigger, R13) both need.
- R13 Phone-side tap-to-trigger prototype for moment cues S Needs Dylan Opus high S19
Cut point: Phone-side debug mode: tapping a word near a tagged moment fires its stinger over the bed, using the time-based position estimate + R12's map — no gaze hardware involved. Dylan judges whether the moment-cue feel is worth building gaze precision for.
Waiting on:
R12· Renderer char-offset<->screen-rect map ·SC11· score-timeline schema extension: point/moment cues ·SN18· Stinger asset family + one-shot mixer layerMore
Amendment (Dylan, 2026-09-23): judge the feel on the phone, cheaply, before any gaze hardware — G10 (armed-cue gaze mode, hw) depends on a pass here.
Score pipeline (Phase 4)
Verified 7.1% · 14 rows
The offline preprocessor: parse, tag, merge into scene cues, validate.
- 1 verified on device
- 5 built, not device-verified
- 8 planned
- SC1 Score timeline model + strict loader/validator + Frankenstein seed (Gutenberg #84) M
Cut point: Timeline JSON validates; `cueAt(book)` unit-tested; golden asset test green.
More
Evidence:
1c0c8de · lib/timeline/score_timeline.dartProved by: 8 tests + golden asset · 2026-06-30
- SC2 Score conductor: reading position → cue → mixer through the MixSink seam M
Cut point: First cue commits immediately; later cues obey the 1200 ms hysteresis; unit-testable without SoLoud.
More
Evidence:
1c0c8de · lib/timeline/score_conductor.dartProved by: 3 tests; emulator init chain · 2026-06-30
- SC3 Scored reader screen (session owner: StemBank → mixer → timeline → conductor) M
Cut point: Reading Frankenstein changes the mix live.
More
Evidence:
1c0c8de · lib/app/scored_reader_screen.dartProved by: emulator; S25 pending · 2026-06-30
- SC13 DCC score timeline: scene cues over the owned EPUB, chapter + char-offset anchored L Needs Dylan Opus high S3
Cut point: A score-timeline.json for DCC book 1 validates against the schema, with per-scene cues using the ST10 roles; it lives outside git and loads with the book.
More
Tagged in-session (as Frankenstein’s 80 cues were), not by the SC5 ClaudeTagger, which stays gated on SN14. The timeline is derived from copyrighted text: gitignored, never pushed.
Evidence:
tool/book_text.py + tool/gen_book_timeline.py -> private/dcc1/ (gitignored); test/private_score_timeline_test.dartProved by: 237 scene cues over 47 chapters + epilogue, validated by the builder and by the app parser against dcc_11m; pushed to the S25 and played · 2026-09-23
- SC14 DCC v2 timeline: place + feel per beat, about twice the beats, the place layer steady L Needs Dylan Opus high S5
Cut point: The book is re-read with a two-axis brief; every cue carries one environment and one feel bed; the environment changes only when the place does; the timeline validates in Python and in the app parser.
Waiting on:
ST14· DCC v2 bed library on Eleven Music: ~32 beds, third- and tonal-centre-gatedMore
Shorter cues were Dylan's other ask; the steady place layer is what keeps denser feel changes from flapping the mix. Dylan, S25 2026-09-24, after S4: the score is clean "but still not with the emotion of what I'm reading"; getting the feeling right per passage is the project's hardest task. So S5 judges feel-fit, not just repetition: Dylan rates passages blind for how well the music fits.
Evidence:
tool/gen_book_timeline.py (two layers, roles.MIX, stand-ins) · private/dcc1/dcc2.score-timeline.jsonProved by: 476 cues, 2.0x v1; the place changes 72 times; test/private_score_timeline_test.dart green. The S25 read is split out as SC15. · 2026-09-24
- SC15 DCC v2.1 heard on the S25: place-first, quiet, not repetitive, and fitting the passage S Needs Dylan Opus high S6
Cut point: Dylan reads DCC on the S25 with the v2.1 pack (ST15, ST16, SN22 in) and no longer calls it repetitive or distracting; a blind bench re-run beats 2.92.
Waiting on:
SC14· DCC v2 timeline: place + feel per beat, about twice the beats, the place layer steady ·ST15· DCC v2.1 place beds: 20 pure field-recording ambiences on Stable Audio 3.0 ·ST16· DCC feel beds that fit: system, show, defiance, respite and guide re-briefed and re-made ·SN22· Moment cues: one-shot sounds fired on their sentence (a blast, a door shaking)More
Split from SC14 (S5): the timeline is built; the device verdict is this row.
- SC4 preprocessor/ Python subproject: EPUB parse → paragraphs with chapter + char offsets M Ultra review Ultracode Opus xhigh S9
Cut point: Frankenstein parses into a paragraph table whose offsets match the reader's chapter geometry.
Waiting on:
SN14· Phase 1 exit: 45-60 min sustained read without loop fatigueMore
Hard-gated on the Phase 1 sonic verdict (review point, 2026-07-04).
- SC5 ModelTagger interface + ClaudeTagger (Opus 4.8 default, Batch API + prompt caching) M Ultra review Ultracode Opus xhigh S9
Cut point: One chapter tagged {mood, intensity, tempo, confidence} per paragraph; cost logged.
Waiting on:
SN14· Phase 1 exit: 45-60 min sustained read without loop fatigue - SC6 Strict schema validation of the generated cue JSON S Opus xhigh S9
Cut point: Generated output passes the same validator the app loads with.
- SC7 Smoothing + merge of paragraph tags into per-scene cues M Ultracode Opus xhigh S10
Cut point: Auto cue count and boundaries are in the same order of magnitude as the hand 80-cue timeline; diff reviewed.
Waiting on:
SC5· ModelTagger interface + ClaudeTagger (Opus 4.8 default, Batch API + prompt caching) - SC8 Stem + transition assignment and full-timeline validator M Ultracode Opus xhigh S10
Cut point: A complete auto Frankenstein score-timeline.json loads in-app and drives the mixer on the emulator.
- SC10 Model taste-test: Opus 4.8 vs Haiku 4.5 vs Jev on golden chapters — agreement, calibration, cost, latency L Needs Dylan Opus high S11
Cut point: Default tagger chosen and written into PREPROCESSOR.md; best auto timeline listened to on device.
Waiting on:
SC8· Stem + transition assignment and full-timeline validator - SC11 score-timeline schema extension: point/moment cues S Opus high S6
Cut point: score-timeline.json schema extended with a point/moment cue type (chapter + char-offset, one-shot stinger reference, no duration); validator updated; documented in SCORE_TIMELINE_SCHEMA.md as planned-then-implemented.
More
Evidence:
docs/SCORE_TIMELINE_SCHEMA.md (Moment cues), lib/timeline/score_timeline.dartProved by: built with SN22 (S6): the moments array, validated in Dart and Python · 2026-09-24
- SC12 Preprocessor moment tagging (sparingly) M Ultracode Opus xhigh S19
Cut point: Preprocessor tags moment candidates sparingly (a handful per book, not per paragraph) on a golden chapter; over-tagging flagged as a defect, not a feature.
Waiting on:
SC11· score-timeline schema extension: point/moment cues ·SC5· ModelTagger interface + ClaudeTagger (Opus 4.8 default, Batch API + prompt caching)
H1 finish line (Phase 5)
Verified 0% · 7 rows
Seed library, comfort settings, the whole-book read that closes H1.
- 7 planned
- P1 Two more public-domain books (Gutendex) scored end-to-end M Opus + Sonnet high S12
Cut point: Two new timelines generated, validated, packaged as assets.
Waiting on:
SC8· Stem + transition assignment and full-timeline validator - P2 Library screen: pick a book, resume position M Opus + Sonnet high S12
Cut point: Three scored books selectable; position restored per book.
- P3 Master volume + mood-intensity control S Opus + Sonnet high S13
Cut point: Both persist and apply live.
- P4 Sleep timer (fade out, stop) S Opus + Sonnet high S13
Cut point: Timer fades the mix and stops the engine.
- P5 Crossfade + hysteresis tuning knobs (debug) with sane defaults S Opus + Sonnet high S13
Cut point: Defaults chosen from the S4 verdict; knobs hidden behind the debug toggle.
- P6 Persisted comfort settings + first-run defaults S Opus + Sonnet high S13
Cut point: Font, theme, volume survive a restart; first run opens on a sensible page.
- P7 H1 exit: Dylan reads a whole book with Libretto and reaches for it again L Needs Dylan Opus high S15
Cut point: Verdict recorded in VISION/BUILD_LOG; H1 declared done.
Waiting on:
SN14· Phase 1 exit: 45-60 min sustained read without loop fatigue
Build, security & platforms
Verified 50% · 10 rows
Forks, git-LFS, OGG, scripted-content sandbox, iOS.
- 5 verified on device
- 2 built, not device-verified
- 3 planned
- PL1 Flutter toolchain + Android app skeleton (es.tinkr.libretto) M
Cut point: Empty app builds, runs on the Pixel 7 emulator, `flutter test` green.
More
Evidence:
ccee4f8Proved by: emulator run + test · 2026-06-22
- PL2 Vendored fork: flutter_inappwebview_android (AGP 9 proguard fix) M
Cut point: Debug + release APK build on AGP 9.0.1.
More
Kept after the S1 drop check (PL7); the two-line delta and the next check date (2026-12-20) are in third_party/flutter_inappwebview_android/FORK.md.
Evidence:
b29ed63 · third_party/flutter_inappwebview_androidProved by: APK builds; runs on device · 2026-06-23
- PL3 Vendored fork: file_picker 11.0.2 (AGP 9 + builtInKotlin=false Kotlin fix) S
Cut point: Release APK builds with file_picker on AGP 9.
More
Fork DELETED 2026-09-20 (S1, PL7): upstream file_picker 13.1.0 federated the Android plugin and fixed the AGP-9 Kotlin guard. Row kept so nobody re-files it.
Evidence:
0704cc3 (fork) · dropped in PL7Proved by: APK builds · 2026-07-04
- PL4 INTERNET permission in the main manifest (release-only WebView spinner fixed) S
Cut point: Release APK renders the reader (no eternal spinner).
More
Evidence:
3aa8aceProved by: emulator, release build · 2026-07-04
- PL5 Audio-master policy: masters out of git (no LFS) + a pre-commit budget guard S Opus + Sonnet high S1
Cut point: A 300 MB WAV drop cannot produce a bad commit; the rule is in CLAUDE.md.
More
No device dimension: the guard is repo-side and was verified by an actual refused commit. LFS refused because it makes a bad commit cheap, not impossible, and bills the owner for every clone of a public repo.
Evidence:
tool/check_audio_budget.py, hooks/pre-commit, docs/AUDIO_ASSETS.mdProved by: hooks/pre-commit rejected a real 13 MB 24-bit/96 kHz drop + a stray .mp3; 16 rule tests · 2026-09-20
- PL6 Shipping format decided: OGG Vorbis q4, with a placeholder/placeholder_ogg format A/B in the app S Opus + Sonnet high S1
Cut point: Decision written (WAV master / OGG shipped or WAV only) with the APK-size numbers behind it.
More
The q4-vs-WAV listening call is SN17, not this row; fallback q6. Flutter assets are STORED in the APK (1.00x), so WAV costs full PCM. "iOS won’t do OGG" does not apply: flutter_soloud compiles libvorbis in.
Evidence:
docs/AUDIO_ASSETS.md, assets/stems/ogg/, tool/check_loop_seam.pyProved by: WAV 7.49 MB vs OGG q4 0.78 MB on a hard 90 s bed; +0 frame drift round-trip; APK 74.5 MB builds green · 2026-09-20
- PL7 Fork drop check: file_picker fork dropped (upstream 13.1.0), inappwebview fork kept with a FORK.md S Opus + Sonnet high S1
Cut point: Either a fork is dropped, or FORK.md records why not and the next check date.
More
file_picker went federated at 12.0.0 under a new maintainer; android_file_picker 2.0.0 handles builtInKotlin=false. inappwebview kept: its fix exists only in a 7-month-old 1.2.0-beta.3 that would drag a beta platform-interface under the reader. Next check 2026-12-20. Open-from-device rides the next APK that reaches Dylan (SN12) - the picker crossed a major version.
Evidence:
pubspec.yaml, lib/app/home_screen.dart, third_party/flutter_inappwebview_android/FORK.mdProved by: release APK builds green on AGP 9 without the fork; analyze clean, 41 Dart tests · 2026-09-20
- PL10 Per-ABI APK split for the sideload build S Opus + Sonnet high S13
Cut point: The APK Dylan sideloads is roughly halved with no asset change and still installs on the S25.
More
Found while measuring PL6: the 74.5 MB release APK is 53 MB of THREE ABIs of native libs and only ~19.6 MB of assets. --split-per-abi (or --target-platform android-arm64) is the whole fix.
- PL8 Scripted-content sandbox for user-supplied EPUBs L Ultra review Ultracode Opus xhigh S13
Cut point: A hostile EPUB fixture cannot run script; user books still render.
More
HIGH before any stranger loads a book: allowScriptedContent:true today, and CA1 (S12) puts friends and family on it. RE-SCHEDULED 2026-09-23 (hardware-first pivot) from the shelved H2 run into H1 S11, ahead of CA1, not deferred — the surface it protects is exposed at S12 regardless of the pivot.
- PL9 iOS platform: flutter create --platforms ios, signing, TestFlight build L Needs Dylan Opus high S102
Cut point: A TestFlight build installs on an iPhone; WAV/OGG loops verified on iOS.
More
Needs an Apple developer account and a cloud Mac.
Books, licences & partners
Verified 0% · 5 rows
Public-domain sources, licensing partners, the legal edges of "score your own books".
- 5 planned
- C1 Admin Books table fed from the same book manifest the app ships S Opus + Sonnet high S12
Cut point: One manifest, two readers (app + /admin/books).
- C2 Catalog content model: licence, source, provenance, timeline version per title M Opus + Sonnet high S105
Cut point: Every catalog title carries a machine-checked licence + source; Gutendex import script.
- C3 Licensing partner scoping: PublishDrive / Ingram CoreSource / Draft2Digital — deal shape, revenue share, DRM obligations M Needs Dylan Opus high S109
Cut point: A written comparison and one opened conversation.
- C4 Soundstripe app-embedding licence quote (the one stock catalog built for apps) S Needs Dylan Opus high S109
Cut point: Quote in hand; go/no-go against a commission.
- C5 Legal read: format-shifting / personal-use scoring under Costa Rica Law 6683, US DMCA §1201, EU S Needs Dylan
Cut point: A one-page note by a lawyer; user-facing claims adjusted.
More
DRM stripping is never a path; "Open from device" stays scoped to files the user legally holds DRM-free. MOVED 2026-09-23 (hardware-first pivot): earlier, into hw — "Open from device" is a desk-prototype path too, not only an eventual online-scoring one.
Crowdfunding & marketing media
Verified 0% · 10 rows
Device render, media, test-user approvals, the campaign itself.
- 10 planned
- CA1 Friends & family test group installs APK v7 S Needs Dylan Opus high S14
Cut point: At least 3 people reading with Libretto and reporting.
Waiting on:
SN16· APK v7 (library + polish) - CA2 Two books driven to "100% approved" by Dylan and the test group L Needs Dylan Opus high S14
Cut point: Fix list → fixes → re-approval; both books signed off.
More
Campaign gate, half 1.
- CA3 Industrial-design brief for the reader device (form, display, camera/IR placement, audio) M Needs Dylan Opus + Sonnet high S310
Cut point: One-page brief approved by Dylan.
Waiting on:
D10· Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policy ·D14· Pi+HAT+battery+speaker in a printed PETG shell, off-battery form/fitMore
RE-RUN 2026-09-23 (hardware-first pivot): moved from h1/S13 into hw, off real e-ink + enclosure learnings (D10/D14) instead of the webcam gaze spike alone.
- CA4 3D render / mockup of the device L Needs Dylan Opus + Sonnet high S310
Cut point: Render set (front, side, in-hand) approved by Dylan.
Waiting on:
D10· Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policy ·D14· Pi+HAT+battery+speaker in a printed PETG shell, off-battery form/fitMore
Campaign gate, half 2. RE-RUN 2026-09-23: moved from h1/S13 into hw, dependent on the real hardware (D10/D14) rather than freestanding.
- CA5 Marketing media: stills + a short video of the hardware prototype and the render M Needs Dylan
Cut point: Media folder ready for a campaign page and social posts.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on batteryMore
RE-RUN 2026-09-23 (hardware-first pivot): retargeted from app growth to hardware production; moved into pcb, unscheduled until pcb sessions are numbered, dependsOn D21.
- CA6 Landing section on libretto.tinkr.es (public-facing, not the roadmap) S Opus + Sonnet high S16
Cut point: A visitor understands the product in 20 seconds.
- CA7 Campaign page draft + reward tiers M Needs Dylan
Cut point: Kickstarter draft with tiers, goal sized to hardware production (tooling + certification), not a stem pack.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on batteryMore
RE-RUN 2026-09-23: retargeted to hardware production; moved into pcb, unscheduled, dependsOn D21 (a session-14 row cannot depend on a session-400+ row).
- CA8 Artist letters: author-recommended artists shortlist + outreach template M Needs Dylan Opus high S17
Cut point: Five names with a one-page pitch each; rights reality (sync + master) explained in the letter.
- CA9 Campaign goal + timeline decision S Needs Dylan
Cut point: Goal, launch window (post-D21) and fallback written down.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on batteryMore
RE-RUN 2026-09-23: retargeted to hardware production; moved into pcb, unscheduled, dependsOn D21.
- CA10 Crowdfunding campaign launch L Needs Dylan
Cut point: Campaign live once D21 passes, funding the production run with the approved books and render.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery ·CA4· 3D render / mockup of the device ·CA7· Campaign page draft + reward tiers ·CA9· Campaign goal + timeline decision ·PU9· Paid attorney trademark clearance + filing decisionMore
RE-RUN 2026-09-23: was run h2 (public-app beta launch); h2 is now shelved, so this moved into pcb, dependsOn D21 in place of CA2/CA4/CA7.
H2 public app
Verified 0% · 9 rows
Store, accounts, payments, onboarding, legal, launch.
- 9 planned
- PU1 Accounts + sync (positions, library) L Ultra review Opus xhigh S104
Cut point: Sign in on two devices; position and library follow.
- PU2 Catalog + scored packs (public-domain seed library, downloadable timelines) M Opus + Sonnet high S105
Cut point: Catalog lists ≥ 10 scored public-domain books.
Waiting on:
P2· Library screen: pick a book, resume position - PU3 Payments + "score your own books" subscription (Play Billing / App Store / Stripe web) L Needs Dylan Ultra review Ultracode Opus max S106
Cut point: A test purchase unlocks scoring; refunds + cancellation paths proven.
Waiting on:
O3· Online mode: "score as you read" chapter-ahead tagging with offline fallback + app mode switch ·PU1· Accounts + sync (positions, library) - PU4 Onboarding flow M Opus + Sonnet high S107
Cut point: A stranger reaches a scored page within 60 s of install.
- PU5 Store assets (icon, screenshots, listing copy) S Needs Dylan Opus + Sonnet high S107
Cut point: Assets approved by Dylan.
- PU6 Privacy policy, terms, support channel M Needs Dylan Opus + Sonnet high S107
Cut point: Published pages; support inbox answered.
- PU7 Public beta + launch checklist M Needs Dylan Opus high S110
Cut point: Play internal test → open beta; checklist all green.
Waiting on:
PL8· Scripted-content sandbox for user-supplied EPUBs ·PU1· Accounts + sync (positions, library) - PU8 Free trademark search: TSDR, TMview, OEPM, UK IPO, WIPO Global Brand Database S Needs Dylan
Cut point: All five free registers searched (~20 min); results logged, including the Dragonfly Games Class-9 "Libretto" word-game hit and Prestigio's EU e-ink-reader use.
More
REDEFINED 2026-09-23 (hardware-first pivot): split off the free search (this row, no dependency, this month) from the paid clearance + filing (PU9, gated at the pcb-horizon exit D21). Don't ship the bare word alone on the physical device — see VISION.md.
- PU9 Paid attorney trademark clearance + filing decision S Needs Dylan
Cut point: Attorney clearance search complete; filing decision made and logged, gated at the pcb-horizon exit — before CA7/CA9/CA10 go public, or before ordering packaging with the name printed on it.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery
H2 online mode & scoring service
Verified 0% · 3 rows
"Score as you read" for unscored books; the preprocessor as a service.
- 3 planned
- O1 JevTagger: second ModelTagger over TypeSafe Jev (Choice on the mood enum, Score for intensity) M Needs Dylan Opus high S11
Cut point: Tags a golden chapter through the Jev API with calibrated confidence per paragraph.
Waiting on:
SC5· ModelTagger interface + ClaudeTagger (Opus 4.8 default, Batch API + prompt caching)More
Needs Dylan's Jev waitlist key. Cloud-only, text-only; never in the live reading loop.
- O2 Scoring service: the preprocessor as an authenticated, rate-limited VPS endpoint with a job queue L Ultra review Ultracode Opus xhigh S103
Cut point: An EPUB posted to the endpoint returns a validated timeline; abuse limits proven.
Waiting on:
SC8· Stem + transition assignment and full-timeline validator - O3 Online mode: "score as you read" chapter-ahead tagging with offline fallback + app mode switch L Ultra review Ultracode Opus xhigh S103
Cut point: An unscored EPUB gets a score while reading; airplane mode still reads with the default bed.
Waiting on:
O2· Scoring service: the preprocessor as an authenticated, rate-limited VPS endpoint with a job queue
Gaze tracking
Verified 0% · 12 rows
Phone-camera spike (h1) -> dark-room IR + Arducam rig re-measured on real hardware (hw) -> custom IR module on the PCB (pcb). Darkness and glasses are first-class; the gate before hardware money.
- 12 planned
- G1 Camera-gaze smoke test: MediaPipe iris → regression → coarse gaze → throwaway CameraGazePositionSource feeding ScoreConductor L Ultracode Opus xhigh S7
Cut point: Gaze drives a cue change end-to-end on video, on the S25 or a laptop webcam.
More
Moved up from H2.5 on 2026-09-20: cheap, no purchase, shapes the device render. Frames never leave the device.
- G2 Gaze accuracy log + go/no-go note for the device concept S Opus xhigh S7
Cut point: A 2-minute real-reading session, hand-labeled, scored on three tiers — line-transition recall ≥80%/≤20% false-positive, region-level (correct third of the page) ≥75%, stretch line-level ±1 ≥60% — and gaze must beat the time-based ReadingPositionSource on the same session.
More
Lit room only, webcam pipeline (~2-4° error): region-level is the realistic bar, line-level is not expected here — that is what G9 tests on IR hardware in the dark.
- G8 Patent/FTO skim: gaze-based reading-position tracking S Opus high S7
Cut point: Patent/FTO skim on gaze-based reading-position tracking, referencing docs/GOTCHAS.md's "not ownable, don't spend on copyrighting the idea" finding; confirms it still holds now that gaze is the pivot's core mechanism, logged under docs/research.
- G9 Dark-room IR gaze spike: Arducam+IR LEDs, darkness + glasses, before the desk prototype locks a camera L Needs Dylan Ultracode Opus xhigh S7
Cut point: Same tiered bar as G2 (region ≥75%, line-transition ≥80%/≤20% FP, stretch line ±1 ≥60%), re-measured with an Arducam OV9281 (global-shutter, no IR-cut) + 850/940nm IR LEDs rig in near-darkness + a dim front light, with and without glasses (single-vision, AR-coated, progressive); 850nm-vs-940nm (provisional until the G12 sun test) and camera+IR duty-cycle-at-rest choices logged; non-camera alternatives (EOG, near-eye photodiode, mmWave, thermal, event cameras) recorded as considered and rejected for a handheld at 30-45cm.
Waiting on:
G2· Gaze accuracy log + go/no-go note for the device conceptMore
Dylan reads mostly in bed, in the dark: this is the primary use case, not a stretch scenario. Gates G6 (the same rig re-measured on the Pi+HAT hw rig) — the hw stage does not lock a camera choice before this passes.
- G3 Gaze + page-estimate fusion, calibration flow, confidence gating L Ultracode Opus max S305
Cut point: Fused position beats the page estimate alone on a 20-minute read; falls back cleanly in bad light.
Waiting on:
G1· Camera-gaze smoke test: MediaPipe iris → regression → coarse gaze → throwaway CameraGazePositionSource feeding ScoreConductor - G4 On-device reading-behaviour heuristics (skim / re-read / pause) → intensity nudges M Opus high S306
Cut point: Nudges are audible, subtle, and computed entirely on-device.
Waiting on:
G3· Gaze + page-estimate fusion, calibration flow, confidence gating - G5 Verdict sitting: does gaze → score feel magical on real hardware? M Needs Dylan Opus high S307
Cut point: Verdict written; feeds the D21 pcb-horizon go/no-go (does not replace it).
Waiting on:
G4· On-device reading-behaviour heuristics (skim / re-read / pause) → intensity nudgesMore
RE-RUN 2026-09-23: moved from the retired h25 run; doneWhen no longer names "the H3 capital gate" — that gate is now D21, at the end of the pcb horizon.
- G6 Gaze accuracy re-measured with the Arducam+IR rig on the Pi+HAT rig M Opus xhigh S308
Cut point: Same tiered bar as G2/G9 (region ≥75%, line-transition ≥80%/≤20% FP), re-measured with the Arducam+IR rig on the Pi+HAT rig at real reading distance; glasses pass included; gaze still beats the time-based ReadingPositionSource on the same session.
Waiting on:
G2· Gaze accuracy log + go/no-go note for the device concept ·D10· Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policy ·G9· Dark-room IR gaze spike: Arducam+IR LEDs, darkness + glasses, before the desk prototype locks a camera ·G12· Sunlight IR gaze test: park/beach midday, 940nm + bandpass + pulsed differencing, before the camera locks - G10 Armed-cue gaze mode: idle-to-armed ramp, fail-silent, 150ms word-level bar M Ultracode Opus xhigh S311
Cut point: The tracker idles at low rate, ramps to 30-60Hz within a few lines of a marked moment, and fires a stinger only on a confident fixation on the target word. Fail-silent verified (low confidence skips the moment, never guesses); word-level bar measured as the share of armed moments firing within ~150ms of the first fixation on the target word, with zero early fires across the test session.
Waiting on:
G6· Gaze accuracy re-measured with the Arducam+IR rig on the Pi+HAT rig ·R13· Phone-side tap-to-trigger prototype for moment cuesMore
Amendment (Dylan, 2026-09-23): early is worse than late (a stinger before the word spoils the scare), so a skipped moment always beats a guessed one. Prototyped cheaply first at R13 (phone, touch, no gaze hardware) before this row commits gaze-rig time to it.
- G11 Dozing detection: eyes-closed fade-out and position save S Opus high S311
Cut point: Eyes closed for N seconds fades the soundtrack out and the reading position is remembered; N tuned so a normal blink or paragraph re-read never triggers it.
Waiting on:
G6· Gaze accuracy re-measured with the Arducam+IR rig on the Pi+HAT rigMore
Amendment (Dylan, 2026-09-23): most relevant exactly where the IR rig is strongest — reading in bed, in the dark.
- G7 IR gaze module on the custom board, IEC 62471 Exempt Group documented M Needs Dylan Ultra review Opus xhigh S404
Cut point: IR gaze module integrated on the custom board; LED choice, drive current, optics and duty cycle documented against IEC 62471, targeting Exempt Group — the same classification Face ID, Windows Hello and all-day Tobii accessibility trackers sit in — including the G12 pulsed outdoor mode.
Waiting on:
D18· First integrated custom PCB rev fabbed and bring-up tested ·G6· Gaze accuracy re-measured with the Arducam+IR rig on the Pi+HAT rigMore
Amendment (Dylan, 2026-09-23): eye safety is a compliance step, not a blocker — the design TARGET is Exempt Group by LED choice/current/optics/duty-cycle, phrased and measured that way, not as an open-ended hazard question.
- G12 Sunlight IR gaze test: park/beach midday, 940nm + bandpass + pulsed differencing, before the camera locks L Needs Dylan Ultracode Opus xhigh S20
Cut point: G9's tiered bar (region ≥75%, line-transition ≥80%/≤20% FP) re-measured outdoors in direct midday sun, open shade and a bright park/beach setting, with the rig hardened for sun: 940nm LEDs + a matched narrow bandpass filter, short global-shutter exposure with LED pulses synced to it, LED-on/LED-off frame differencing to subtract ambient IR; with and without glasses; sunglasses tested and, where they block IR, detected with a clean fallback. When confidence drops (glare, squint, sun on the lens) position falls back to the time-based ReadingPositionSource and moment cues go fail-silent, verified on the same session. Final 850-vs-940nm choice logged with both dark (G9) and sun evidence.
Waiting on:
G9· Dark-room IR gaze spike: Arducam+IR LEDs, darkness + glasses, before the desk prototype locks a cameraMore
Amendment (Dylan, 2026-09-23): e-ink is at its best outdoors, so reading in a sunny park or at the beach is a real use case. Sunlight is the hard case for IR gaze (it floods the sensor with its own IR and shrinks the pupil); 940nm sits in a water-absorption dip of the solar spectrum, which is why outdoor trackers prefer it. Gates G6: the hw rig does not lock a camera or wavelength before both dark and sun pass. The pulsed mode raises LED peak power, so G7's IEC 62471 check must cover it.
Device
Verified 0% · 25 rows
The reading device: desk prototype (hw) -> custom PCB + enclosure (pcb) -> manufacturing (h3 epics, reachable once D21 passes).
- 23 planned
- 2 dropped
- D11 Real APK sideloaded on a borrowed AOSP e-ink device (OS-answer test) M Needs Dylan Opus high S300
Cut point: Real APK sideloaded on any cheap/borrowed AOSP e-ink device; settles Android/AOSP vs. a custom Flutter e-ink embedder; no gaze hardware or G2 required.
More
Folds the retired D1 (same test, dropped 2026-09-23 — see D1).
- D24 Open-hardware e-ink reader build write-ups read end to end (PineNote, Open Book Touch, Modos Flow) S Opus high S300
Cut point: All three build write-ups read end to end; lessons logged to docs/research.
- D9 Pi 5 + flutter-pi: flutter_soloud mixer proven off-phone L Ultracode Opus xhigh S301
Cut point: The flutter_soloud mixer plays a real crossfade through BT/speaker on a Pi 5 running flutter-pi, off the phone, zero changes to lib/mixer.
Waiting on:
P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it againMore
Nothing in hw that costs real money starts before P7 closes — schema-enforced via this dependency.
- D10 Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policy M Ultracode Opus xhigh S302
Cut point: Own script pushes frames to the Waveshare HAT over USB/SPI; full- vs. partial-refresh latency logged; a refresh-cadence policy picked.
Waiting on:
P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it again - D12 Combined-load check: audio + e-ink refresh proven simultaneously on one Pi 5 S Opus high S303
Cut point: Audio and e-ink refresh proven running simultaneously on one Pi 5; SPI/CPU/power contention logged.
Waiting on:
D9· Pi 5 + flutter-pi: flutter_soloud mixer proven off-phone ·D10· Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policyMore
The row a first draft of this plan left out entirely — nothing else verifies the two halves coexist before D14's enclosure or D2's OS decision.
- D13 Phone app pairs with the Pi rig over BT/USB S Opus high S303
Cut point: Phone app pairs with the Pi rig over BT/USB; phone can see the device's current book/position.
Waiting on:
D9· Pi 5 + flutter-pi: flutter_soloud mixer proven off-phoneMore
The first real deliverable behind the companion-app narrative: the phone app becomes the soundtrack lab + companion once a device exists.
- D14 Pi+HAT+battery+speaker in a printed PETG shell, off-battery form/fit M Needs Dylan Opus high S304
Cut point: Pi+HAT+battery+speaker close inside a printed PETG shell, run off the Stage-2 LiPo pack; form/fit only.
Waiting on:
D9· Pi 5 + flutter-pi: flutter_soloud mixer proven off-phone ·D10· Waveshare IT8951 e-Paper HAT driven directly, refresh cadence policy ·P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it againMore
Low-priority idea (Build Vault 4a801b14, 2026-10-06): for the shell mounting plates and brackets, Claude writes a parametric CadQuery/FreeCAD script directly (big model writes the spec, local CAD builds it). Not Taiga-S1: the tiny HF text-to-FreeCAD model dropped holes and looped on corner fillets in its own demo. Nothing installed yet.
- D2 Android/Boox-sideload vs. custom Flutter e-ink embedder, ratified L Needs Dylan Opus high S304
Cut point: Decision memo with power, cost and vision-stack trade-offs, ratified against D11's sideload results; prototypes stay on Android until then.
Waiting on:
D11· Real APK sideloaded on a borrowed AOSP e-ink device (OS-answer test)More
RE-RUN 2026-09-23 (hardware-first pivot): moved out of the h3 epic bucket and shrunk XL → L now that it has a real cut point; dependsOn D11 alone (corrected from D9/D10/D11) so a gaze "no" can never accidentally stall the OS decision the whole stack needs.
- D25 Real APK + gaze rig on a purchased Boox tablet (gaze-on-hardware sideload) M Needs Dylan Opus high S309
Cut point: Real APK + gaze rig on a purchased Boox tablet; full chapter reads with the score changing live and gaze assisting position, screenshots/video captured.
Waiting on:
G2· Gaze accuracy log + go/no-go note for the device concept ·P7· H1 exit: Dylan reads a whole book with Libretto and reaches for it againMore
Gated on G2 (a gaze "no" defers Tier C/D spend, not the whole hardware track) and on P7, because the Boox is real money like every other hw purchase (Codex review, 2026-09-23).
- D15 First KiCad board routed and submitted to fab M Needs Dylan Opus high S400
Cut point: A simple 2-layer board routed in KiCad, ERC/DRC clean, submitted to JLCPCB.
- D17 PMIC/rail-sequencing literacy spike S Needs Dylan Opus high S400
Cut point: Rail-sequencing diagram for the chosen panel exists, checked against the vendor datasheet.
- D20 Good Display/E Ink contacted for the waveform file + hobbyist pricing S Needs Dylan Opus high S400
Cut point: Good Display/E Ink emailed re: waveform file + hobbyist-quantity pricing; reply or non-reply logged.
- D16 Charger + fuel-gauge bring-up on the desk prototype M Needs Dylan Opus high S401
Cut point: BQ24074 + MAX17048 wired, SOC reporting over I2C on the desk prototype.
Waiting on:
D9· Pi 5 + flutter-pi: flutter_soloud mixer proven off-phone - D18 First integrated custom PCB rev fabbed and bring-up tested L Needs Dylan Ultra review Ultracode Opus xhigh S402
Cut point: First integrated custom PCB rev (fuel gauge + charger + IT8951 breakout) fabbed and bring-up tested, using D20's waveform file and D17's rail plan.
Waiting on:
D15· First KiCad board routed and submitted to fab ·D16· Charger + fuel-gauge bring-up on the desk prototype ·D17· PMIC/rail-sequencing literacy spike ·D20· Good Display/E Ink contacted for the waveform file + hobbyist pricingMore
Real money onto an expensive panel for the first time — sequenced (not left to session-numbering luck) so power is never wired blind.
- D19 Enclosure CAD v1, dimensioned to the real board+panel+battery M Needs Dylan Opus high S403
Cut point: Enclosure CAD v1, dimensioned to the real board+panel+battery, SLA-printed.
Waiting on:
D14· Pi+HAT+battery+speaker in a printed PETG shell, off-battery form/fit ·D18· First integrated custom PCB rev fabbed and bring-up testedMore
Same CAD route as D14: Claude-written parametric CadQuery/FreeCAD script, not Taiga-S1 (Build Vault 4a801b14).
- D21 PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery L Needs Dylan Ultra review Opus max S405
Cut point: A hand-built unit — custom PCB rev 1 + SLA/FDM case — runs a full reading session, with gaze assist, on battery.
Waiting on:
D18· First integrated custom PCB rev fabbed and bring-up tested ·D19· Enclosure CAD v1, dimensioned to the real board+panel+battery ·G7· IR gaze module on the custom board, IEC 62471 Exempt Group documented - D1 Android e-ink reader as test device (Boox Go 7 / Palma 2, or Hisense A9 for a front camera) + APK on e-ink M
Cut point: Libretto reads on e-ink with acceptable refresh and audio out.
More
DROPPED 2026-09-23 (hardware-first pivot): folded into D11, the redefined OS-answer sideload — same test, broader device pool (any AOSP e-ink device, not only Boox Go 7/Palma 2/Hisense A9). Kindle jailbreak ruled out for Flutter (2026-09-20): flutter-pi needs DRM/KMS.
- D3 Device prototype: IR gaze module, display, audio, battery XL
Cut point: A bench prototype reads a book with gaze-driven score.
More
DROPPED 2026-09-23 (hardware-first pivot): obsolete as a single epic — split across D9 (audio), D10 (e-ink), D16 (charger/fuel-gauge), G6 (gaze rig re-measured on real hardware), G7 (IR module + IEC 62471 on the custom board). Ground truth for that earlier research: Pupil Labs Core (~3.6k EUR) or Gazepoint GP3 HD (~2.7k USD); consumer Tobii is not licensed for gaze data.
- D4 ODM RFQs (RK3566-class e-ink boards) + open-hardware references (Open Book Touch, Modos Flow) XL Needs Dylan
Cut point: Three quotes with MOQ and NRE; one reference board on the bench.
Waiting on:
D2· Android/Boox-sideload vs. custom Flutter e-ink embedder, ratified - D5 Word-level cue anchoring (gaze precision → cue timing on the dot) XL Needs Dylan
Cut point: Cues fire within one line of the intended word on the shipped device.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on batteryMore
The armed-cue mechanism (fail-silent, ~150ms bar) is already prototyped at G10 (hw) and re-verified at G7/D21 (pcb); this epic carries it onto the manufactured run.
- D6 On-device reflex classifier over gaze features (only if a local System-One / open-weight model exists) XL Needs Dylan
Cut point: A local model classifies reading state under 20 ms on the device; no frame leaves it.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on batteryMore
Conditional epic from the Jev research: cloud reflex models are the wrong shape for this loop.
- D7 Funding (crowdfunding proceeds, loan, or investment) for a production run XL Needs Dylan
Cut point: Capital secured for the first run.
Waiting on:
CA10· Crowdfunding campaign launch ·D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery - D8 Manufacturing + fulfilment of the first run XL Needs Dylan
Cut point: Devices in backers' hands.
Waiting on:
D7· Funding (crowdfunding proceeds, loan, or investment) for a production runMore
MOQ asymmetry named explicitly: boards scale down cheaply (50-500 units), the molded case does not (500-1,000). Certification: FCC Part 15C pre-cert radio module (cheap) vs custom RF; EU CE-RED plus the 2026 Cyber Resilience Act cybersecurity gate, which can turn a ~1,500 EUR self-assessment into a 60,000-120,000+ EUR Notified Body program — see D22.
- D22 FCC pre-cert module chosen; EU CRA scope-and-cost read M Needs Dylan
Cut point: FCC Part-15C pre-cert module chosen; EU CRA scope-and-cost read.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery - D23 Small-batch contract manufacturers vetted, RFQs sent M Needs Dylan
Cut point: 2-3 small-batch CMs vetted, RFQs sent.
Waiting on:
D21· PCB-horizon exit: hand-built unit reads a full session, gaze-assisted, on battery