Home / Ops / More in this area

The Attuned Reading: Structure Spec v1 (the demo's #2 delivery, the A-job replacement for the chart-pure Mirror in the demo sequence)

Updated Jul 18, 2026 · Affirmology_AttunedReading_Structure_v1.md

The Attuned Reading: Structure Spec v1 (the demo's #2 delivery, the A-job replacement for the chart-pure Mirror in the demo sequence)

Written 2026-07-18 by Cowork per Jeff's PP-1 decisions of the same day. This is the definitive structure for the second gift in the demo week: the CONTEXT + CHART, ACTIONABLE reading that lands ~day 2-3 via the live Mirror-drip. It builds directly on Affirmology_FirstReading_Structure_v1.md (mirror_recognition_v1) and inherits everything there that is not explicitly changed here. Craft sources: Affirmology_ReadingCraft_HeroStory_Synthesis_v1.md, Affirmology_AhHaMoves_Registry_v1.md, Affirmology_ContextCapture_AttunedReading_Concept_v1.docx. No em dashes.

Naming. Customer-facing name stays THE MIRROR (already announced in email + hub; the structure underneath changes, the name does not). Internal structure id: attuned_recognition_v1, a SIBLING of mirror_recognition_v1, never a context-injected fork of it.

The one-line brief. It leads chart-first and gets MORE specific than the Origin did (more specificity is what kills "too generic"), names the listener's doubt out loud and reverses it, then turns to the life area they chose on the results page and pays it off with concrete, actionable substance. Chart leads, context follows. The actionability is the whole upgrade over the B-job Mirror.

Sourcing rule (unchanged). Composes from the person's Deep Map or a fresh focused council pass, never raw chart JSON. Style shared, content never shared. All generation through the council layer.

0. The three input tiers (what the renderer receives)

The results-page box gives us up to two things: one tapped life area and one optional filtered sentence. The structure must degrade gracefully across all three states. One structure, three fills; never three different structures.

The joke filter's silent-fallback rule holds: a troll or nonsense entry renders as T2, and nothing about the reading acknowledges the rejected text. Nothing sensitive is ever surfaced that they did not knowingly give.

The shape at a glance

12 to 14 minutes, ~1,900 to 2,200 words at Storyteller pace (the drip render path's length band). One story circle. TWO chart doors instead of three (the minutes they give up fund the area session). ONE master symbol threaded start to finish. One loop always open. The doubt-reversal beat lives in the first ninety seconds. The crescendo lands inside the area session, not before it. Spoken, no music bed, slightly slower pace, real pauses.

Word budget by block: Return ~180, Deepening ~260, Door One ~420, Door Two ~420, Turn ~120, Area Session ~600, Install + Seal ~250.

Stage map (engine stage words in parentheses)

1. THE RETURN, 0:00-1:15 (opening)

• Re-entry as a continuing story, never a cold open: "When we last spoke, I read you from your sky alone. You have been on my mind since."
• THE DOUBT-REVERSAL BEAT, the signature move of this structure. Name the doubt before they can: "And maybe, somewhere in the days after, a small voice said: that could have been anyone. Good. I want that voice here for this, because in the next few minutes it is going to run out of places to stand."
• The promise, structured as a two-part information gap: first, "I am going to get more specific than last time, specific enough that the doubting voice goes quiet." Second, name their chosen area WITHOUT opening it: "and then we are going to take everything we find straight into [their area], because you told me that is where life is loudest right now. That part I am saving."
• T3 variant: replace the second promise with the salience lens named as OUR choice: "and then we are going into the one conversation your whole chart keeps having, because it would be malpractice to leave you without it."
• Plant the MASTER LOOP: the thing in the chart that "changes what [their area] has been asking of you." The loop and the area are fused; curiosity now points at the payoff section.
• Craft retrieval: opening query (hook promise curiosity gap doubt inoculation specificity).

2. THE DEEPENING, 1:15-3:00 (opening into induction)

• The falsifiable strike: ONE precise inner experience from the Deep Map's felt-pattern bank, a claim a stranger would not risk. "You count the exits in every room" grade. This is the direct answer to the doubting voice, delivered within two minutes of naming it.
• The receipt line, spoken once and lightly: "A horoscope could not have said that. Your chart just did."
• First sacred pause with breath instruction.
• Introduce the MASTER SYMBOL from the person's metaphor bank. Selection rule for this structure: prefer a symbol that can WORK in their chosen area (a keel can hold a marriage steady; a lamp can light a career). The symbol must be able to travel into the area session and do labor there, not just decorate the chart half.
• Voice enters low, warm, intimate.
• Craft retrieval: induction query (prosody pause parasocial descend warmth).

3. THE TWO DOORS, 3:00-7:30 (journey_structure)

Two doors, not three. Selection is salience-scored AND area-weighted: among the top convergences, prefer the two with the strongest natural bridge into the chosen area (the Wanted signal finally has real weight, we asked and they answered). Each door runs the full segment template with a Vonnegut shape line, ~2 to 2.5 minutes:

1. Felt pattern first, multi-domain examples.
2. The reveal with the convergence refrain: "Four systems, built centuries apart. Same address."
3. The mechanism as a causal chain, with fear preemption where needed.
4. Contrast engine: two roads, both true, neither shamed.
5. The portable law + the practical move.
6. The affirmation line, spoken twice, slower the second time.

Door assignments: - Door One, 3:00-5:15: the loudest convergence. The core signature, the "how did it know" hit. The master symbol does its first full work here. This door may glance toward the chosen area in one clause, never more; the area is still being saved. - Door Two, 5:15-7:30: the juiciest tension. Shadow-to-gift, opened with the polarity conundrum rehook. Somatic checkpoint at the deepest point. This door should end angled toward the area: the shadow named here is the same force they will meet again inside the area session, and the script may say so as the seam rehook ("keep your hand on that, because it is about to matter where you live"). - Retention layer unchanged: one loop always open, rehook at every seam, ethical variable reward. - Craft retrieval: journey query (story circle open loop contrast shadow gift symbol convergence).

4. THE TURN, 7:30-8:15 (journey_structure)

• The hinge of the whole structure, and the intake-mismatch move's home. Three sub-beats, ~40 seconds total: 1. Acknowledge the choice as THEIR act: "You chose [area]. One tap. I want you to know your chart heard it." 2. The convergence claim that fuses context and chart: "Here is what you could not have known when you tapped it: everything we just walked through has been about [area] the whole time." 3. T1 only: ONE brief echo of the structured extraction, reframed in OUR language, maximum one clause, never their verbatim sentence. Example shape: they wrote about money stress despite doing fine; the script says "you told me safety and money do not sit in the same room for you yet." That is the entire allowance. The reading must never become a mirror of their own words; their words bought the aim, the chart supplies every round. 4. Mismatch variant: when the chart is genuinely shouting about something other than the chosen area, use aha move #12 and bridge: "You asked about love, and your sky keeps answering about what you are building. Stay with me, because those are the same conversation." Never override their choice; always land inside the area they picked.
• Craft retrieval: turn query (bridge mismatch reframe chosen question thesis).

5. THE AREA SESSION, 8:15-11:45 (journey_structure)

The payoff and the reason this structure exists. The engineered peak of the week; the crescendo lives here. It runs an extended segment template with the actionability ladder replacing the single "practical move" slot.

• The pattern in the area, 8:15-9:15. The chart's testimony about their chosen area, pattern-first: how the Door One signature and the Door Two tension actually BEHAVE in this specific territory. Multi-slot specifics inside the area (for relationships: the first week of knowing someone, the fight that repeats, the silence after; for wealth codes: the invoice not sent, the price not raised, the generous streak that empties the tank) so they find their own furniture. T1 aim sharpens which slots lead.
• The mechanism and the two roads, 9:15-10:15. The causal chain of the pattern inside the area, then the contrast engine walked in area-specific terms. THE CRESCENDO lands at the top of the high road: the single most important identity line of the reading, fullest delivery, warmth over volume. Mid-block, touch the master loop once ("we are almost at the piece I saved").
• The actionability ladder, 10:15-11:45. The concrete substance the Mirror feedback asked for. Three rungs, all three always present, each anchored to the chart so instruction never floats free of identity: 1. The law. One portable, quotable principle for the area, derived from the chart mechanism. ("You are not built to receive by asking. You are built to receive by being findable.") 2. The practice. One repeatable protocol they could run weekly, stated in literal steps a listener can hold without paper: when, what, how long. Concrete nouns, real times of day, an ordinary-day anchor ("Tuesday morning" grade). 3. The move. ONE tiny act for THIS WEEK, Fogg-tiny, doable in under five minutes, framed as an identity vote: "before Sunday, [literal act]. Not to fix anything. Because that is what a person with your chart does when she remembers who she is."
• The ladder ends with the area affirmation, spoken twice, slower the second time. First person, present tense, area-specific.
• Hope-dominant framing is law here, especially for inner child and wealth-fear contexts: non-catastrophic, non-clinical, the existing reading rules apply in full. The dip is earned and never stranded.
• Craft retrieval: area query (actionable prescription tiny habit identity vote lifestyle ordinary day + the area name).

6. THE INSTALL AND THE SEAL, 11:45-13:30 (closing)

• PAY OFF THE MASTER LOOP: the saved thing, the piece that reframes the two doors AND the area as one story. It must specifically recast the area: after this line, their chosen area should look different than it did when they tapped it.
• The Ericksonian install, identity-framed and area-anchored: "as you begin to notice how this already shows up in the small moments of [area]... you are someone whose X comes from the inside now."
• Bookend the opening promise explicitly, including the doubting voice: "Somewhere back at the start I invited a small voice along. Notice how quiet it has gotten."
• The three affirmation lines restated once, breath-paced (Door One, Door Two, Area).
• Close INSIDE the master symbol, doing area work in its final image.
• One deliberate forward loop pointed at the third gift, small and warm, never a hard CTA: "there is one more thing I want to make for you. It is different from anything you have heard from me yet. A few days."
• Settle into calm certainty.
• Craft retrieval: close query (closing affirmation cadence identity install symbol payoff forward loop).

Hard rules (the QC gate for this structure)

Rules 1 through 12 of mirror_recognition_v1 all still apply (no checklist walk, no naked jargon, one master symbol, loops paid, two rehooks minimum, one crescendo, law + move + twice-spoken affirmation per door, one somatic checkpoint minimum, composer-seam strip, word band now 1,850-2,250, earned dips, no fear bait, listener is the hero). New rules for this sibling:

13. Chart leads, context follows, mechanically enforced. The chosen area may be NAMED before the Turn exactly twice: once in the opening promise (the saved-for-later loop) and once as Door Two's seam rehook. It may never be OPENED there: no area content, no area-aimed specificity, no free-text reference of any kind before the Turn. Reject any script whose first falsifiable strike or first door aims at the chosen area. The first seven minutes must stand alone as chart dazzle.
14. The one-clause echo cap. The person's free-text context may surface at most ONCE, at most one clause, always rephrased into our language, never their verbatim sentence. Reject any script that quotes the user back to themselves or that references their text more than once. (T2/T3 scripts must contain zero implied text references.)
15. The doubt-reversal beat is mandatory and must complete inside the first 90 seconds: doubt named, doubt invited to stay, first falsifiable strike delivered by 3:00. Reject scripts that skip the inoculation or that answer the doubt with a generic claim.
16. The actionability ladder is mandatory and complete. All three rungs (law, practice, move), each chart-anchored, the practice with literal steps and a real-world time anchor, the move sized under five minutes with a this-week boundary. Reject a script whose area section is insight-only. This rule is the point of the A-job.
17. The area section gets the minutes. Area session no shorter than 3:00 of the runtime. Reject scripts that spend the area in a coda.
18. The crescendo lives in the area session, on the identity line at the top of the high road. Reject scripts that peak in the doors and coast through the payoff.
19. Graceful tier degradation. A T3 render must read as a complete, intended reading, never as a form with empty fields. Reject any script containing conditional-sounding seams ("since you did not tell us...").
20. Never punish the skip and never shame the doubt. The doubting voice is welcomed, not scolded. The box-skipper gets full magic.

The QC pass (how to run the gate)

Before render: verify the script against rules 1-20 as a written audit (the same audit-before-render flow as mirror_recognition_v1). Machine-checkable first (word band, echo count, area-section length, ladder presence, first-context-mention position), then the craft judgment calls. Log the structure id + tier + audit result per person (structure feedback loop, Atlas metrics), so tier-specific failures become visible fast.

Test plan

1. Generate all three tiers for the SAME test person (Sol as calibration listener, per the standing plan): a T1 with a real sentence, a T2 area-only, a T3 skip. The three-way read is the fastest way to see tier seams.
2. Email Jeff the scripts + the written 20-rule audit before any render.
3. Render the T1, deep-listen in the car, iterate.
4. Then a small live batch through the drip on new demo-takers, with the tier logged, watching: open rate on the #2 email, completion, the "needs action" note (it should disappear), and any "it just repeated what I typed" note (rule 14 failure, the one new way this structure can die).

Handoff notes for Code

• Renderer selects attuned_recognition_v1 for every demo #2 delivery once live; mirror_recognition_v1 no longer renders in the demo sequence (Jeff, 2026-07-18). The chart-pure Mirror can live elsewhere in the app later.
• Inputs consumed: Profile.current_context (the parse/filter pass's structured extraction + the tapped area + a tier flag). The prompt scaffold should receive tier explicitly, not infer it.
• The drip needs no changes: same queue, same gates, same email-to-page delivery. Length band update only (12-14 min).
• The parse/filter pass should emit: area (enum), context_extract (one clean sentence, our phrasing, or null), usable (bool). The reading never sees raw user text.