Affirmology - Test & Change Queue (the single forced list)
Updated Jun 24, 2026 · Affirmology_TestAndChange_Queue_v1.md
Summary. Jeff asked for "some way to access, be told about, and forced into a list of tests and changes." This is it: the ONE living list of what to test (verify it works) and what to change/build, prioritized, each with the EXACT next action. Kept current by Cowork; s
Affirmology - Test & Change Queue (the single forced list)
Jeff asked for "some way to access, be told about, and forced into a list of tests and changes." This is it: the ONE living list of what to test (verify it works) and what to change/build, prioritized, each with the EXACT next action. Kept current by Cowork; surfaced to Jeff on a schedule (see the morning-digest scheduled task). Check items off as done.
Priority: P0 = blocking / do now. P1 = high value, soon. P2 = important, queued. P3 = later.
NOW (P0, blocking everything downstream)
[x] PUSH THE HERMES LOOP FIX + RESTART THE SERVER. DONE - loop fix is live on cloud; the two-person wealth read returned a full reading on studio.affirmology.ai 2026-06-20.
[x] HARDEN HERMES (P1) - DONE + DEPLOYED 2026-06-20 (commits ca3a055 + eb6c075). Front-end now polls /api/hermes/history (no more "could not reach my model" flash); residual multi-oracle BadRequestError fixed (forced-synthesis keeps tools=). Live on studio.affirmology.ai (app.js v=19).
[x] R2 CRAFT SYNC - DONE 2026-06-20. craft (94) + techniques (18) merged into the R2 corpus; verified live on cloud /api/corpus-stats (craft=94, techniques=18). Cloud Hermes/--council now use the real craft base.
[ ] HERMES CHAT ACCURACY GATE (P1): Apollo's craft-critique gate only runs on --council audio renders, NOT on Hermes chat readings, so chat-side factual slips (e.g. a 2026-06-20 reading conflated Pearl 44.1 with 45.5) go uncaught. CLOSE IT TWO WAYS: (1) feed each slip into the growth engine feedback stage so it becomes a permanent distinction; (2) add a lightweight accuracy self-check to the Hermes loop for chart facts (gate identities, which gate is the Pearl, house/sign placements) before it answers. To spec.
NOTE: the residual multi-oracle BadRequestError seen from the funnel/Cowork path on 2026-06-20 is the LOCAL server running pre-Fix-B code; the local restart clears it (cloud already has Fix B). Escalate only if it persists after a restart.
TESTS TO RUN (verify it works) - after the loop fix is live
[x] T1 [P0]: WEALTH READING - PASSED 2026-06-20. Two-person (Jeff+Sol) wealth-codes read returned a full mechanism-level reading (Pearls 45.5/49.5, channels, Lilith/Chiron blocks, live transits), not a stub. Loop + oracles confirmed working on cloud.
[ ] T2 [P1]: TIMING. Ask Hermes "what money opportunities are opening for me right now?" and "what big thing happened around my late twenties?" PASS = it cites current transits / the Saturn return / progressed season from the timeline.
[ ] T3 [P1]: FULL BLUEPRINT IN A PULL. Ask Hermes to read your chart. PASS = the answer can reference aspects, the Lots, Chiron, the asteroids (the engine now carries them).
[ ] T4 [P2]: DEMO NUMEROLOGY. Confirm the demo no longer shows name-based numbers (Expression/Soul Urge/Personality), only date-based.
[ ] T5 [P2]: PWA ON IPHONE. Add Atlas to your home screen; confirm the icon, the Hermes/Studio toggle, cross-device chat history, and the mic (speak-to-text) work.
[x] T6 [P2]: GROWTH CRON - PASSED 2026-06-20. The affirmology-growth-nightly cron fired on schedule (09:00 UTC = 5am ET), ran the three growth stages, and pushed an enriched corpus back to R2. After a web-service redeploy, cloud /api/corpus-stats shows meaning=20, distinctions=1 (size grew to 18,874,368). Runs daily at 5am ET.
CHANGES / BUILDS TO DO
[x] C1 [P1]: RENDER NIGHTLY CRON - DONE 2026-06-20. affirmology-growth-nightly is live (schedule 0 9 * * * = 5am ET), 6 secrets saved. First run 2026-06-20 produced growth (meaning=20, distinctions=1); see T6.
[~] C2 [P1]: CRAFT KNOWLEDGE ENGINE. Phase 1 DONE (94 craft entries seeded). Phase 2 DONE + DEPLOYED 2026-06-20 (wired into the council: Hypnos plans/sequences techniques from the live base, Orpheus drafts with the plan, Apollo gates on it, Hermes search_craft tool in chat; static TECHNIQUES_LIBRARY now fallback only). Cloud deployed AND local funnel server restarted 2026-06-20 (running latest build 2026-06-20-send-digest-resend, craft=94 live locally). REMAINING: Phase 3 (nightly crawler grows the craft tradition) + Phase 4 (frequency/sound layer in audio_mix.py, = C3).
[x] C3 DONE + DEPLOYED 2026-06-23 (studio commit 13becb2 pushed to origin/main). FREQUENCY / SOUND LAYER. Synthesized tone bed UNDER the voice+music mix, state-matched (delta 2 / theta 6 / alpha 10 / beta 14 Hz) via Structure.bed_profile; isochronic default (speaker-safe) + binaural option (headphone-flagged); new agents/tones.py (state map + honesty note + tones_enabled) + audio_mix.synthesize_tone_bed/overlay_tone_bed/apply_tone_layer; cli --tones (OFF by default) / --tones-modality / --tones-volume; the SAME audio_qc.py runs on the final tone-bearing mix (fails -> drops to the no-tone mix); tone meta + honesty note written to the manifest + build report; gk_single_sphere bed_profile -> theta_soft (journey theta). Studio-only, additive, OFF by default, demo byte-for-byte unaffected, 1.05 ceiling + QC preserved. 5/5 acceptance + full studio suite 74 pass (test_c3_tones.py). LISTEN-CHECK delivered to Jeff (delta/theta/tones-off, tone applied to the existing deep mix, $0 extra ElevenLabs, both QC pass). NEXT: Jeff decides which structures turn tones on (the flag is off until then); optional Studio UI toggle + a fresh full journey render later. Binaural / solfeggio / isochronic / tonal overlays in audio_mix.py, structure-aware (delta for sleep, theta for journey), Studio-only, demo locked. Spec: Affirmology_FrequencySoundLayer_C3_Spec_v1.md.
[ ] C4 [P2]: MUSIC BEDS (Jeff is behind, blocking audio production). Generate more beds in Suno from the ready prompts in Affirmology_SunoBedPrompts_v1.md (Sleep/Journey/Morning/Gym), log keepers in Affirmology_MusicAudition_Log_v1.md, then they slot into the music-bed selector. ACTION: open the Suno prompts doc, generate, log.
[ ] C5 [P2]: MUSIC-BED SELECTOR (finish). Wire the selector + the pace categories (the data's there; needs the backend --music + render test). Code job.
[ ] C6 [P2]: TECHNIQUES -> CORPUS. Fold the Techniques Library into a searchable "techniques/craft" tradition (folded into C2 above).
[~] C7 [P2]: VEDIC DEPTH. Vimshottari dasha, divisional charts (D2/D9/D10), Dhana yogas (for the wealth read). Engine build. Consolidated build brief: Affirmology_VedicDepth_C7_BuildBrief_v1.md (supersedes the older CLAUDE OUTPUTS/Affirmology/Affirmology_VedicEngine_PhasedPlan_v1.md). PHASE L1 (Vimshottari dasha timeline) STARTED 2026-06-21 (handed to Code). Remaining: L2 vargas+yogas, L3 Vedic corpus, L4 Agastya + Vedic audio paths. Fixture-first: nothing ships until it matches a trusted Vedic calculator.
[ ] C8 [P2]: DOMAIN PLAYBOOKS. Build the relationships playbook next (Eros + lenses, strong coverage), then career (Athena).
[ ] C13 [P3]: NATIVE APP WRAPPER (Capacitor) when you want the fully-native feel + on-device Parakeet voice. NOW PROMOTED via the END-USER APP V1 (below).
[ ] APP-V1 [P1, NEW TRACK]: END-USER APP V1 (player-first). The consumer app: receive your Sacred Audios + a premium PLAYER that LOOPS, plays in BACKGROUND (keeps going when you leave the app / screen off), and has lock-screen + AirPods controls + a sleep timer. PWA can't do iOS background audio, so v1 = Capacitor native wrapper (reuses the web app, adds native audio). GATE: Jeff gets the Apple Developer Program (~$99/yr) to build/TestFlight/ship. Spec: Affirmology_EndUserApp_v1_Spec_v1.md. Also covers the library + onboarding + tier/credits. Open scope Qs in the spec.
[ ] C30 [P1, audio system]: LAYERED STEM BED ENGINE (Option B). A stems/ library per state (drone/pad/texture/pulse/motif, tagged by key+bpm+role in manifest.json) + a bed-assembler in audio_mix.py (same stage as C3 tones) that picks a palette by bed_profile, loops/stretches to the script's section markers + target length, automates layer gain across the arc, ducks under voice, adds the C3 tone tuned to the palette key. Fed by Suno ISOLATED-ELEMENT generation (ambient states) + Demucs/Suno stem splitting (rhythmic states). Graceful fallback to the single preset bed; demo locked. Depends on C26 + C3. Spec: Affirmology_LayeredStemEngine_OptionB_Spec_v1.md. Build order: C26 -> C3 -> C30.
[ ] C14 [P3]: WISDOM CIRCLE VISUALS (your parallel art track: Oshun, Hildegard, an Asian/Polynesian priestess, Sol Ballard from photos).
[ ] C31 [P0, CRAFT RULE - Jeff 2026-06-24]: NEGATION + SHADOW-NAVIGATION RULE. Council-confirmed (Chiron + craft corpus): present-tense negations backfire. In theta the limbic/somatic network loads the content word ("arriving", "broke", "behind") before the prefrontal "not" lands (often it does not land at all); ironic-process theory + our Believable-Gap entry confirm it. RULE: (1) NEVER use present-tense negations or name the unwanted state in the installation body; name only the state being built. (2) If contrast is needed, past-tense the old state ("I used to...") then land fully in the present-tense new state with >=2 declarative beats; do the pattern-interrupt with a PAUSE in delivery, not a "not" in the words. (3) Witnessing of the struggle lives in the induction/somatic-settling layer, not the install body. SHADOW WORK IS KEPT (Jeff): shadow->gift->siddhi journeys and the TWO shadow expressions (repressive + reactive poles) stay core (Persephone's lane); word the shadow as a ROUTE the listener navigates and reroutes through, not a present-tense state they inhabit, and always land present-tense in the gift/siddhi. Same reroute move for anything previously phrased as a negative. IMPLEMENT: (a) add the rule to script generation (Orpheus draft directive) + an Apollo gate that flags present-tense negations in the install body; (b) Chiron + Persephone formalize the navigational shadow-phrasing patterns (two-pole reroute language) as a craft-corpus entry; (c) light daytime/activation tracks stay build-only by TYPE. Rule doc: Affirmology_NegationRule_Craft_v1.md.
NEW FEATURE REQUESTS (from Hermes chat, 2026-06-20) - to scope
[ ] C15 [P3, DEFERRED - this is the LOW-priority casual one, NOT C18]: CASUAL READ-ALOUD / FIELD BRIEFING AUDIO (a quick voiced readout of a chat conversation, "tell me about that abundance thing while I make lunch"). Lightweight TTS, not a Sacred Audio (that is C18, HIGH). Jeff 2026-06-20: "that can wait." SCOPING: cost + delay per render; delivery model (one-tap "send to Studio" that pre-populates the script vs a background render that drops into a library tab); free vs premium tier. TTS ENGINE: Jeff flagged Fish Audio as a candidate for the fast/cheap voice path (ElevenLabs stays for full Sacred quality). Demo stays locked; this is a Studio/Atlas capability. DELIVERY (Jeff 2026-06-20): Hermes renders the audio and plays it INLINE in the chat, then asks "save to Studio as a shareable audio?" (-> C17 persist, C16 share). Cost-controlled: tier-gated + HARD per-member server-side cap, OpenAI tts-1 default (no new account). Build brief: Affirmology_FieldBriefingAudio_C15_BuildBrief_v1.md. Build AFTER C16/C17 (reuses their plumbing); public share LINK needs C10 (R2 public media). DEPRIORITIZED 2026-06-20 (Jeff: not an immediate need). ENGINE DECISION when built: Fish Audio hosted (~$0.10/audio, no hardware) is the pick; ElevenLabs too costly (~$0.50/audio); do NOT stand up the old Mac mini for local TTS (the cloud-hosted API removes any hardware dependency). Tier-gated + hard per-user cap still apply. PRIORITY: P3 / later.
[x] C16 [P1] - DONE + LIVE 2026-06-20 (see the C16 + C17 built line below). CROSS-CHAT MESSAGING VIA HERMES ("the messenger god delivers"). One member pushes a reading, note, timing report, or field briefing from their Atlas chat into another named member's chat (Jeff <-> Sol <-> advisors; later coach <-> client). On-brand internal comms layer built on the chart + reading infrastructure. Pairs directly with C15 (push a field-briefing audio as the payload). THIS RESOLVES the previously-deferred "Atlas in-app messaging" question: Jeff now clearly wants it, Hermes-side, cross-chat push. Permission tiers from the Atlas spec apply (Jeff full; Sol/Colin reduced). GREENLIT TO BUILD 2026-06-20. Build brief written: Affirmology_CrossChatMessaging_C16_BuildBrief_v1.md. ACTION: paste the C16 Code brief to the studio Code window.
[x] C16 + C17 BUILT + LIVE 2026-06-20 (commit 96d8dd6; prompt fix 849d9e1 so Hermes actually knows the new tools). C16 cross-chat messaging (send_to_person, /api/hermes/push|inbox|inbox/read, deliveries table, chat buttons + inbox) and C17 send-to-Studio (send_to_studio creates a Studio DRAFT entry, /api/studio/draft + /api/studio/draft/{id}/render). All acceptance tests pass; permissions enforced server-side. Cloud auto-deployed; LOCAL needs a restart to pick up the code + prompt fix + the Phase 2 craft engine.
[x] C18 [P0, HIGH - Jeff 2026-06-20 "100%, exactly what I want"] - DONE + DEPLOYED 2026-06-20 (core commit 4a0e0ea; follow-ups commit 788c358: deliver_to + "Cast" button). Confirmed live (build tag in running cloud + local server). HERMES-RENDERED SACRED AUDIO. The real priority: Hermes takes the chat reading/script and triggers a FULL Sacred Audio render through the EXISTING proven pipeline, then delivers the finished audio + PDF back in chat. No manual Studio step. The /api/studio/draft/{id}/render endpoint ALREADY EXISTS (Code, live); remaining work = let Hermes TRIGGER it (a tool/flow on user request) + surface the finished media + a render-status update in the chat. Plus: after delivery, Hermes prompts to make a SHARED record searchable by all Studio users (opt-in, additive, server-side permission). Uses ElevenLabs / the real pipeline (core product, no new cost analysis needed). Demo untouched. NEEDS A NAME for the capability (Cast/Invoke/Conjure/Summon, Jeff to pick). Build brief: Affirmology_HermesRenderedSacredAudio_C18_BuildBrief_v1.md. CORE BUILT + LIVE 2026-06-20 (commit 4a0e0ea, build 2026-06-20-c18, app.js v21): render kickoff -> completion delivery (audio+PDF) in chat -> share to a searchable Shared Library; permissions pass. NAME = Cast (chosen). FOLLOW-UPS pending: (1) deliver_to param so Hermes can Cast an audio into ANOTHER circle member's chat (Jeff->Sol), reusing C16; (2) label the UI button "Cast." Code paste delivered 2026-06-20.
PRIORITY CORRECTION (Jeff 2026-06-20, "get it straight"): HIGH = Hermes renders full SACRED AUDIOS from chat (C18). DEFERRED / P3 = casual read-aloud TTS of chat convos "while making lunch" (the old C15 field-briefing/cost-memo feature, below). They are different features; do not conflate.
NOTE: C16/C17/C18 are ONE action-layer family ("Hermes acts across the toggle"); the endpoint+tool+button pattern is built, C18 completes it for the Sacred Audio render.
[x] C19 [P1, efficiency] - DONE + DEPLOYED 2026-06-20 (commit 8452cda, build tag 2026-06-20-c19-stored-natal-basis; engine pytest 46/46, C19 acceptance 8/8). STORED NATAL CHART BASIS (Jeff 2026-06-20). Stop recomputing the immutable natal chart on every audio/reading. Compute once, store (keyed by birth-data hash + CHART_ENGINE_VERSION so engine upgrades recompute once), reuse as the basis for all Sacred Audios, readings, analysis, Hermes context. Timing (transits/progressions) stays on-demand. Precompute the 4 known people as committed fixtures. Big compute saving at scale + consistency + speed. Build brief: Affirmology_StoredNatalChartBasis_C19_BuildBrief_v1.md. Sequence AFTER the urgent BadRequestError fix + the C18 follow-ups land.
[ ] C20 [STRATEGIC / future, multi-phase]: THE LIVING PROFILE (Customer Intelligence Layer). Jeff's vision 2026-06-20: deeper per-lens chart analyses (prosperity/relationship/career/health/growth/spirit) + assessment + behavior = a rich per-person profile that drives onboarding, membership tracks, communication voice, content recs, and marketing (email/SMS, chart-aware timing), plus base-wide segmentation/trends. Built ON the C19 stored basis; reuses the council + growth engine. Strategic moat (day-one psychographics). Phased 0-4. Privacy/ethics non-negotiable. Concept doc: Affirmology_LivingProfile_CustomerIntelligence_Concept_v1.md.
[x] C28 [P1] DONE + COMMITTED 2026-06-21 (studio commits a21fbcf nightly lane, e922b6e +28, 8ce25f8 +35 with scope/form, 612797c +18 hypnotherapy researched first + Hypnos wiring, 3975330 rule+form+scope written onto the craft columns). practitioners.py carries teacher, lineage, core_method, mechanism (evidenced/traditional/metaphor), practice, how-it-could-serve-an-audio, tags, confidence, scope, form, rule; copyright tier-walled, public-domain ingested fuller, quantum framing flagged as metaphor, cost-capped; test_practitioners.py green. PRACTITIONERS & TECHNIQUES RESEARCH LANE (nightly). Jeff 2026-06-20: have the nightly builder research transformation/manifestation TEACHERS + techniques (Ken Wilber, Gurdjieff, Neville Goddard, "wave function from the outcome backwards," Jung, Coue, Erickson, NLP, Maltz, law-of-assumption family, etc.), distill into a KEPT + assessed wisdom/practitioners craft-corpus area (teacher, method, mechanism honestly stated, actionable practice, how it serves an audio, usefulness score). Tier-wall copyright; flag quantum framings as metaphor. Extends the craft-engine Phase 3 cron. Seed list + plan: Affirmology_PractitionersTechniques_ResearchCorpus_Plan_v1.md. Jeff to share his ChatGPT "teach me" chats to fold in his curation. PENDING: ~40 more candidates surfaced in a Cowork exploration round (esoteric, meaning/flow psych, trauma/somatic, sound/journey, biology-of-belief, shamanic, stoic/taoist, performance) awaiting Jeff's add-all/trim; plus optional crawler self-expansion (auto-propose adjacent teachers, flagged for review).
[x] C29 SCHEMA DONE + COMMITTED 2026-06-21 (studio commit 347bc09: craft table gains rule (1-10, multi) + form + scope in both engine and vendored craft.py, with the additive migration, search_craft filtering by rule/form/scope, craft_seed emitting rule+form, and the dry-run-safe craft_backfill; test_craft_rule_form.py green). Backfill --apply run 2026-06-21 on the live corpus (112 entries, 106 already tagged, last 6 processed, $0.0038). Journaling prompts + mindset practices drafted in Affirmology_BeliefInstallation_C29_Kickoff_v1.md. C29 [P0, CORE CRAFT SPINE]: BELIEF INSTALLATION mechanism (Jeff's "most important": how the subconscious accepts a belief through repetition + identity WITHOUT rejecting it). 10 rules synthesized (state-first, hypnagogic bypass, small-step ladder, identity-anchored-in-evidence, feeling-as-encoder, first-person present, questions/afformations, pace-and-lead/softeners, spaced repetition + sleep, consistency momentum), each with lineage + evidenced-vs-traditional honesty. Becomes the audio install arc AND the schema that SORTS all research by (which rule it serves) + (which FORM it becomes: audio / journaling prompt / mindset practice / journey companion / breath / sound). Drives app features (journaling + mindset practices wrapped around unlocked journeys). Doc: Affirmology_BeliefInstallation_CoreCraft_v1.md. NEXT: add the form/rule tags to the craft+practitioners schema; draft first journaling prompts + mindset practices for the wealth journey.
[x] C26 DONE + DEPLOYED 2026-06-23 (studio commit 601e7f0 pushed to origin/main; Jeff listen-checked both renders and approved the slow DEEP pacing ~0.88 and ACTIVATION ~1.03 "for daytime"; both audio_qc.py PASS, both under the 1.05 ceiling, demo byte-for-byte unchanged). Studio-only, additive, demo byte-for-byte unaffected (standard pace is a no-op). Structure gains pace + bed_profile (before_bed + gk_single_sphere = deep; confidence_abundance = activation/driving; natal_sacred stays standard). New agents/pacing.py (pace to speed deep 0.88 / standard 0.95 / activation 1.03 under the 1.05 ceiling; mix-time silence after questions + at section breaks; target length per type; deterministic pacing gate). Hypnos sequences pacing, Orpheus writes the question + laddered arc, Apollo gates it; QC + 1.05 ceiling kept. 6/6 C26 tests green, engine 46/46, studio 114 pass (1 pre-existing chart_cache failure, unrelated). NEXT: Jeff listen-checks one deep + one activation render, then STEP 4 (C3 tones). C26 [P1, CRAFT - "make them dynamite"]: AUDIO DELIVERY PACING + ACTIVATION CRAFT (Jeff 2026-06-20). Deep-listening audios (morning, before-bed, journey) need SLOWER delivery, MORE PAUSES (space for words to land), and sometimes QUESTIONS that activate the listener's subconscious (ask, then leave silence). Pace + sound bed vary BY TYPE: deep = slow + spacious + soft bed; "gym"/activation type = same words can move faster, with a different, energizing bed. Implement via: ElevenLabs slower speed for deep types, pause/SSML insertion (and silence after questions), question-craft in the scripts, and per-Structure pace + bed config. NOTE: also fix render LENGTH consistency, Sol's morning audio came out longer than Jeff's; renders should honor a target length per type.
[x] C25 [P0, foundation] - DONE + DEPLOYED 2026-06-20 (commit d028d81, build tag 2026-06-20-personal-knowledge-base). knowledge_base table in studio.db + R2 mirror (survives rebuild), always-injected pinned context into every Hermes turn (survives chat clear + restart), save_to_knowledge_base tool + /api/kb endpoints, Cast brief folds in pinned entries, Jeff + Sol seeded with the standing WEALTH CONTEXT. 39/39 studio tests green. NEXT: UI panel to view/edit the KB; grow toward Living Profile (C20); golden-tier credit model. PERSONAL KNOWLEDGE BASE (persistent per-person standing context). Jeff 2026-06-20: the golden-tier need + the durable fix for context loss. A server-side per-person store (studio.db + R2 mirror) of pinned standing context that is ALWAYS injected into Hermes (independent of chat history, survives restarts + chat clears) and feeds audio creation. save_to_knowledge_base tool + endpoints. SEED Jeff's and Sol's first pinned entry with the standing WEALTH CONTEXT so it is never lost again. Foundation of the Living Profile (C20) + the credit/golden-tier model. Concept + build spec: Affirmology_PersonalKnowledgeBase_GoldenTier_Concept_v1.md. Also: WEALTH SUITE spec for spawning the audio program: Affirmology_WealthBreakthrough_AudioSuite_Spec_v1.md; morning audios written: Affirmology_MorningActivations_JeffAndSol_v1.md.
[x] C24 [P0, CRAFT-CRITICAL] - DONE + DEPLOYED 2026-06-20 (commit 327465c, build tag 2026-06-20-c24-cast-right-script; verbatim plumbing 425dbda). Cast now voices the composed source_script VERBATIM; from-chart Casts get a themed brief (theme + craft directive + pinned KB). 45/45 studio tests green. CAST MUST RENDER THE COMPOSED SCRIPT, NOT "from_chart". Confirmed 2026-06-20: Hermes composes a strong themed first-person script in the chat (wealth/belief/breakthrough, names Affirmology + funding, releases the old timeline), but render_sacred_audio renders generically from the bare chart (source_script defaults to "from_chart") so the AUDIO the user gets is NOT the script shown, it just echoes the natal blueprint. FIX: Hermes must pass its composed script as source_script when Casting, and render_sacred_audio must voice source_script verbatim through the proven render/QC/PDF pipeline; only fall back to from_chart when no script exists. Quality bar: the rendered audio == the script shown in chat, word for word. Benchmark script: Affirmology_BeliefBreakthrough_SacredAudio_Jeff_v1.md.
[ ] C23 [P1]: EMAIL THE SACRED AUDIO. On Cast/render completion, ALSO email the recipient their audio link + PDF (Jeff 2026-06-20: "more records, more memories, more optins") in addition to the in-chat delivery. Reuses the Resend send path (same as the reports), so proving Resend email works once unblocks both. GATE: first confirm a real Resend email actually lands (env vars on Render + a test send). Then wire render-completion -> email.
[x] C22 [P1] - DONE + DEPLOYED 2026-06-20 (commit 7f7809c, app.js v25). Attach/paste .txt/.md/.pdf as read-only context for one question (never stored, never shared); server extracts text (pypdf), per-doc + total caps, injected into the model window only. 9/9 tests. PASTE/UPLOAD FILES INTO HERMES CHAT. Let Jeff attach or paste txt + md files (and best-effort PDFs) into the Atlas/Hermes chat; the server extracts the text and includes it as context so Hermes can read and use the document. Front-end attach/paste + backend /api/hermes accepts attachments + PDF text extraction + size caps. Code paste delivered 2026-06-20.
[ ] C21 [P1, UX]: ATLAS CHAT LINK/MEDIA RENDERING. Hermes replies show raw URLs as plain text (copy-paste, inconvenient). Fix web/app.js to render markdown / auto-linkify (clickable links, new tab) and render audio/PDF deliveries as an inline
[ ] C26 [P1, craft]: BREAKTHROUGH / BELIEF-CHANGE CRAFT TRACK (Jeff 2026-06-20). Deepen the corpus on how to break through blocks and limiting beliefs. Researched what works UNDERNEATH the big names (Tony Robbins NAC, NLP, Byron Katie, EFT, the meditation trend): the real mechanisms are memory reconsolidation (surface -> reactivate -> felt contradiction -> repeat), CBT reframing, identity/values self-affirmation, first-person kinesthetic mental rehearsal/future pacing, state-first physiology, and emotional repetition consolidated in sleep. Plan: new breakthrough craft area, seed via a Haiku distillation, grow via the Phase 3 nightly grower (tier-walled: branded programs benchmark-only), council usage (Persephone owns the arc, Chiron the technique, Hypnos sequences, Apollo gates the reconsolidation sequence), and a Breakthrough technique-slot sequence + structure (Studio only, demo locked). Brief: Affirmology_BreakthroughCraft_BuildBrief_v1.html. NEXT: Jeff approves the area -> I write the seed source notes -> seed + verify -> add structure + render a test.
AUDIO PRODUCTION (Jeff: "I'm so behind on audios")
[ ] A1 [P1]: Generate the music beds (C4 above) so the audios have varied sound.
[ ] A2 [P1]: Churn test audios across the structures (now that they draw on the richer blueprint + craft). Pick 3-5 structures, render, listen, log feedback (which feeds the growth loop).
FROM JEFF + SOL SESSIONS (auto-routed here to get processed)
Source notes get archived under AFFIRMOLOGY and the actionable items land here so the morning digest forces them. Jun 21 2026 Ephemeralogy session notes: AFFIRMOLOGY/Affirmology_EphemeralogyProcess_SessionNotes_2026-06-21_v1.md.
- [ ] S1 [P1]: BRAND NARRATIVE LOCK (Jeff + Sol, Jun 21). Finalize the core story and apply it to the seminar/launch video script + homepage hero. Pillars: (1) the emotional promise is RELIEF ("you were not broken, you were running on a model that was not your blueprint"); (2) tagline candidate "Something always felt like the world was showing you a different way"; (3) the law-of-attraction GUARDRAIL: we are design-aligned, NOT "manifest anything you want", do not over-promise. ACTION: open the notes doc, pull the pillars, draft the launch-video VO + hero copy.
- [ ] S2 [P2]: LOCK THE TWO BRAND PHRASES "soul songs" (meaning/emotional) and "subconscious operating system" (logical/descriptive) into Atlas onboarding copy + marketing. ACTION: add to the brand-language doc and onboarding strings.
- [ ] S3 [P2]: MARKET THE IDENTIFY PHASE AS THE HERO DIFFERENTIATOR. Message that identity work is auto-baked into every audio (Identify -> Align/blueprint -> Transcend), versus other systems that make you do separate identity work and whose affirmations do not land. ACTION: write the "why ours lands" comparison block for the site.
- [ ] S4 [P3]: EXPLORE optional in-audio journaling prompts / self-answered questions inside program audios (ties into the C29 forms: journaling prompt + mindset practice). Decide if it belongs in the program flow. ACTION: spec a small test in one program track.
How this list stays in front of you
A scheduled morning task surfaces the top P0/P1 items each day (the "forced into it" part).
Any new "we should test/change X" from any chat gets added here immediately.
Tell Cowork "mark C4 done" or "what's on my list" anytime.