# ALEVIA × native-SOP audit — CURRENT six Fable stories

## Decision

**All six require targeted structural repair. None earns a literal native-SOP PASS.** This is not a recommendation for six wholesale rewrites. Introductions, recognizable parent/ADHD-child worlds, consequential ordinary incidents, and the broad independence desire now work. The repeated weakness is the middle: the stories earn investigation of a **physical waking cue**, then quietly start describing an unidentified particular band, and only later disclose that it is Dawn. The name's late position does not make that transition earned.

The earlier independent review was right about restored characters and developed imagined endings, but too generous in calling discovery/choice substantially closed. A child asking a better question is the start of discovery, not a complete source-to-decision sequence. A generic comparison and a product claim do not automatically answer the story's central objection.

- **S01:** partial encounter provenance exists through the sister's link; retrieving the deleted recommendation and establishing what page it is are missing transitions.
- **S02/S03/S04/S05/S06:** no functioning exact-brand encounter before an anonymous particular product begins supplying facts. The link at the end identifies the advertised destination, not how the character got there.
- **S02:** strongest gradual confidence arc, but durable-response doubt is not answered by motor strength or a guarantee.
- **S03/S05:** strongest ownership and situational discovery respectively, but their reasons favor a category/form, not demonstrably Dawn over a relevant shortlist.
- **S04:** acknowledges maintenance, yet the central concern—another parent-managed job—needs evidence about the setup/maintenance task, not simply 'no app.'
- **S06:** ordinary work urgency works. Current Joanne name and revised evening-learning/available-backup chronology resolve the prior review's identity and schedule flags.

`matrix.json` is the complete passage-level matrix: **23 functions × six scripts**, each marked present/partial/missing/unsupported, with exact current-file passages and line ranges. Missing and unsupported cells quote their nearest seam or attempted substitute; the quote is not falsely represented as fulfillment. 'Present' means a function works under the authorized adaptation, not that the literal SOP's expert/proof requirements are satisfied.

## 1. Scope, authority and reading coverage

### Current source of truth

Full reads of:

- `/Users/ayden/.openclaw/workspace/output/fable-six-stories-2026-09-10/final/S01.md` through `S06.md`.
- `/Users/ayden/.openclaw/workspace/output/fable-six-stories-2026-09-10/rewrite-brief.md`.
- `/Users/ayden/.openclaw/workspace/output/fable-six-stories-2026-09-10/revised/independent-review.md`.
- `/Users/ayden/.openclaw/workspace/knowledge/formats/zakaria-long-format-native-sop-v2-2026-09-03.txt`, all 712 file lines.
- `/Users/ayden/.openclaw/workspace/output/trendtrack-golden-natives-2026-07-22/teardown-how-they-write-natives.md`, full document; wrong-brand Appendix A excluded from the comparator.

Correct corpus: `/Users/ayden/.openclaw/workspace/output/trendtrack-golden-natives-2026-07-22/corpus-alevia.md`. ALEVIA is alevia.com, page **811974392003013**, brandtracker **9717ee7a-f73f-4cec-b382-10b4ea4aeedb**. It is not aveliacare.com. No short Avelia template was used.

### Three actual complete golden bodies read

| Source ID | Opening | Complete body coverage | Role in this audit |
|---|---|---|---|
| `facebook_1046655904584168` | “I am about to say something that might annoy a few nephrologists, but after what I found out, I can't keep quiet anymore.” | Corpus L13–189, through final CTA; metadata L6–12 also read | Cousin Dave; mechanism, symptom re-audit, detailed juice/powder/capsule sourcing and actual bottle-to-brand bridge |
| `facebook_1397040502290767` | “I was diagnosed with rheumatoid arthritis at 32. I'm 47 now. Turns out I never had arthritis at all.” | L1829–2117, through final root-cause close; metadata L1822–1828 read | PT is simultaneously peer and authority; dialogue, extract/powder comparison, staged micro-actions, clinical validation and economic close |
| `facebook_1899242397445298` | “There's a wooden fishing boat at the end of our pier that hasn't gone out in over a year.” | L4187–4693, through final CTA; metadata L4180–4186 read | Full witness architecture; stranger-to-personal-stakes bridge, search-led rather than expert-led mechanism, sourcing and relationship resolution |

The corpus's initial large read was persisted/truncated in tool display. These three bodies were then deliberately reread in bounded complete ranges; full-body coverage is not claimed for the whole 22-body corpus. August `ad-copies*.json` were unnecessary and were not read. Sources establish what these ads say, **not that their medical claims, testimonials or comparisons are true**. Historical duplication/longevity is not current performance evidence.

Native-ads skill loaded. Its documented Claude counterpart was checked and is absent. The actual named SOP was available in full and controls this audit. No live comments, live destination, hardware tests, customer receipt joins, automated native gates, or CF deployment were performed. This is an authorized closed-scope editorial audit, not live-source substantiation. Notes' cited comments inform objections; they do not become in-story discoveries or results.

### Current-version corrections

Current S04 is **Leah/Evan** (L17–25), current S06 **Joanne/Owen** (L21–35). Do not carry forward duplicate Ben/Claire findings. S04 L201 and S06 L217 no longer use the prior conspicuous analyst formulations. S06 L225–231 now locates learning in an evening and a morning with backup, rather than postponing ordinary work indefinitely. The broader source/selection gaps remain.

The user moved illustration framing outside copy onto the review page. **No missing-primary-disclosure complaint is made, and no reinsertion is requested.** Conditional imagined payoff remains inside all six bodies. Metadata that still tells a publisher to retain a primary illustration label is stale handoff language, not evidence that the externally framed review is secretly a real testimonial. No CF change is authorized here.

## 2. What the originals actually do—sequence, not a universal checklist

### A. Dave/nephrologist: tangible messenger first, brand later, validation last

Actual sequence: provocative confession (L13) → symptom specificity (L15–25) → uncle/dialysis stakes (L27–31) → attempts (L33–37) → reunion contrast and Dave (L39–53) → rust mechanism (L55–67) → each symptom re-explained (L69–73) → distinct failed-solution re-audit (L75–85) → narrator asks what Dave does, bottle appears (L87–91) → fruit skepticism and research (L93–109) → new sourcing objection (L111) → juice, powder, generic capsule comparison (L115–119) → requirements (L121) → **“That is exactly what Dave was holding”** and Alevia (L123–125) → order/use (L127–129) → gradual observations (L131–135) → expectation calibration (L137) → later activity/lab/doctor (L139–149) → other people (L153) → reorder/supply economics/guarantee (L155–165) → explicit two choices (L167–179) → multiple PS re-closes (L181–189).

What transfers: the first bottle encounter is already planted; its later identification resolves an existing object. Each new piece of knowledge triggers the next objection: fruit sounds trivial → investigate compound → juice seems easiest → learn sourcing constraints → identify matching bottle. What does not transfer: supplement medical claims, exclusivity assertions, clinical markers, scarcity, dosage/bundles or unsupported outcomes.

This body does **not** have a credentialed expert deliver the discovery mechanism. Dave does; the nephrologist appears later for validation. Research at L95 is vague, not a named study. It has no full RA-style itemized dollar autopsy. Those are not universal mandatory beats.

### B. Rheumatoid-arthritis/PT: source has two roles; not every comparison precedes naming

Attempts begin before the extended symptom montage (L1833–1855 versus L1857–1871). More attempts follow (L1875–1879). The turning point is a PT stopping an ordinary session and identifying with the narrator (L1885–1905), not a near-fatal incident. Peer and credentialed explainer are the **same person**, not necessarily two successive characters.

Mechanism unfolds through questions and answers (L1907–1959), with symptom re-explanation embedded in the dialogue. The narrator leaves, reads unnamed forums, then texts the **same identifiable person** and receives a lunch-break call (L1961–1969). Ingredient mechanism follows (L1971–1997). The brand the PT used appears with extract/powder differentiators (L1999–2001). Guarantee comes **before** use/results (L2003–2005). Micro-actions and husband observation occur before clinician validation (L2007–2061). A fuller sourcing comparison and stronger failed-treatment re-audit recur **after** results (L2063–2079). Economic autopsy, urgency and stock claims follow (L2081–2107). The final jar callback and CTA have **no PS stack** (L2109–2117).

Transfer: narrator can return to a known source, ask the next practical question, and compare specifics without introducing a fictitious new authority. Do not enforce 'all comparison before brand,' a separate peer-before-expert cast, or mandatory multi-PS closure. Do not import the ad's internally conflicting diagnostic hook as medical truth.

### C. Fishing-boat witness: the teardown materially misdescribes the people

**Walter does not recover. Walter dies.** Jean is his widow; an unnamed narrator meets her and then investigates for her own husband **Mark**, whose recovery the ad claims. The teardown's brief 'wife narrates husband Walter's decline and recovery' gloss cannot be used as the actual plot.

Sequence: observed boat ritual/mystery (L4187–4203) → narrator's cottage and absent Mark (L4205–4211) → early walk, stumble, helping with cooler (L4213–4253) → Jean and Walter's ritual/loss (L4255–4305) → Walter's history, decline and death (L4313–4407) → narrator explicitly maps Walter to Mark (L4409–4455) → exact kitchen-laptop query (L4457–4463) → **narrated research exposition**, not sustained expert conversation (L4465–4509) → attempts re-audited, some first named here (L4511–4519) → ingredient mechanism and unnamed clinical trial (L4521–4563) → sourcing filter and “brand that kept coming up” (L4565–4567) → order/use and staged activity (L4569–4617) → measurement/doctor (L4619–4643) → return to Jean and permission to tell others (L4645–4667) → reader bridge and close (L4669–4681) → PS expectation/guarantee/scarcity re-closes (L4683–4693).

Transfer: mystery becomes relevant through a developed relational bridge; narrator has a specific reason to search. Mark's rod preparation precedes casting, making recovery tangible rather than generic. But this source itself has a weak exact-brand encounter: “kept coming up” names no page. Its exact query does not make unnamed research sources strong. **The user's request for identified pages is a legitimate quality improvement, not a claim every golden body already meets it.** Do not replicate concealed supplementation, unsupported drug claims or death stakes for parent-child independence.

## 3. Three standards that must not be conflated

### Literal SOP

The actual SOP calls for problem-aware 500–800-line scale (§02), expert-led cascade/dialogue (§§03–04/06), acknowledge → gap → summary treatment of alternatives (§07), unique/full-cycle solution and expert recommendation (§§03/08), product at 75–80% (§08), week-by-week results and outside noticing (§09), and formal formatting/process checks (§§10–13). It also requires defensible truth (§03).

All six depart from important literal requirements. They are not literal SOP passes because they have a late brand token and allude to alternatives. Matrix measurements are descriptive primary-text word counts, not line-based or semantic gate passes. Earlier category investigation is not itself forbidden: **sell the requirements before the product**. The problem is anonymous particular-band advocacy, not that a wrist cue is discussed before 'Dawn.'

### Observed ALEVIA functions

Common persuasive work: a reason to keep reading; someone whose life matters; escalating cost of the old approach; information that changes how previous events are understood; a reason the easiest next solution is insufficient; a product connected to those requirements; modest observable moments; confidence from something beyond assertion; a reason to act and a relationship-restoring close.

Order and implementation vary. Narrator naming, direct early reader address, expert dialogue, fully cited research, initial failure stacks, price autopsies and multiple PSs are **not invariant** across the three complete originals. The teardown's 16 beats are a synthesis, not a transcript to enforce indiscriminately.

### Authorized Fable adaptation

Keep current length/detail where a function works; add only the missing causal material. No line-count padding. No requirement to expand to supplement-native length. Keep named mothers/children and existing ADHD context without making age, job, college, moving or diagnosis an eligibility filter. Ordinary independence stakes can replace medical catastrophe. No native prices. A practical cue-delivery explanation can replace biological disease theory; a supported choice can replace exclusive cure. One coherent hypothetical future can replace fabricated retrospective results, but **cannot replace efficacy substantiation**. One PS can resolve the relationship without imitating a five-PS sales stack.

## 4. Seam ledger: knowledge → action → source → fact → objection → evidence → decision

These are missing transitions and repair contracts, not replacement copy. A source identity means a real, inspectable source surface (publisher/page title, actual product/manual/spec page or genuine referral), not necessarily a raw URL stuffed into every paragraph. Do not invent a convenient page or pretend a hypothetical family truly found it.

### S01 — preserve deletion, self-blame and cupboard insight

- **Before:** Rachel thinks differently marketed sound alarms are different approaches; Mia feels recommendations reject her (L86–150).
- **Action/location/query:** cupboard conversation; “Alarms that use vibration” (L112–156). Search is sufficiently directed.
- **Source encountered:** “a page about wrist alarms” (L158); then “the band in the link my sister had sent” (L174). Which page, and how the deleted link was retrieved, are absent.
- **New fact:** wrist delivers vibration to wearer; maker says stronger motor, increasing vibration and on-band setup (L160–188).
- **Objections triggered:** why trust another recommendation; why not ordinary watch; can Mia stop it and remain asleep; why this band?
- **Evidence answer currently:** physical delivery distinction, attributed maker comparison, willingness and dismissibility. No waking evidence; no identifiable source for the motor comparison; guarantee only late.
- **Selection:** “The band was Dawn Band” (L198), then wrist/settings/strength reasons (L200–202). This is the nearest to an encounter but still retrospective naming.
- **P1 contract:** join cupboard question to an identified truthful source; show how sister-link access resumes and why this page gets attention despite recommendation fatigue. Keep maker assertions attributed; answer watch comparison with inspectable facts, not success implications. Let the guarantee address the fear of one more failed purchase without pretending it repairs belief. Acceptance: a reader can recount the source route without supplying a missing click and can distinguish product capability from response uncertainty.

### S02 — preserve empty-bed hook and slowly earned trust

- **Before:** Nina knows occasional success doesn't predict tomorrow; Ella also cannot explain the variance (L97–121).
- **Action/location/query:** table reconstruction → “Is there an alarm you can feel?” → joint search (L123–155).
- **Source:** bed/wrist results → unnamed descriptions → **“This particular band”** (L157–175). No click/page identity.
- **New fact:** escalating vibration; dismissal doesn't require standing; manufacturer motor claim; on-band settings; guarantee (L177–215).
- **Objection:** will it work beyond a hopeful first morning, rather than just can it buzz harder?
- **Evidence answer:** none establishes durable waking. The daughter accepting responsibility and the guarantee legitimately address agency and purchase risk, not durability. This is the main unclosed objection, not an objection-handling pass.
- **Selection:** “That combination made Dawn the one I wanted us to choose” (L217). Combination assembles category/brand features but not a traced encounter.
- **P1 contract:** identify the actual source transition and the limited reason to prefer its design. Distinguish three questions: cue reaches wearer, wearer rises, reliability persists. Answer only supported ones; retain durable response as uncertainty instead of borrowing unrelated comments. Tie trial decision to uncertainty without turning the whole ending back into a test checklist. Acceptance: guarantee cannot be read as proof of consistent waking; late feature cluster no longer arrives from nowhere.

### S03 — preserve child-led invitation and privacy breach

- **Before:** mother assumes setting/moving alarm and withdrawing supervision should suffice; child can set it but misses the wake step (L102–148).
- **Action/location/query:** Ben invites Claire to his desk; laptop research “alarms you can feel” (L102–108).
- **Source:** bed alarm, wrist alarms, “setup information for one of the bands” (L150–158), all unidentified.
- **New fact:** controls on wrist/no app, escalating physical cue, adjustable strap and idle-dark display (L160–186).
- **Objection:** will this preserve child control; why this band rather than another no-app wrist alarm; can he dismiss it and stay in bed?
- **Evidence answer:** settings location and child preference address ownership; explicit dismissibility bounds behavior. No competitor shortlist or unique capability established.
- **Selection:** “The one Ben chose was Dawn Band”; final reason only compares with moving a loud clock (L194–198).
- **P1 contract:** make Ben's research legible through an identified real setup source and a finite actual comparison. Preserve his leadership. Replacing the old clock is a category decision; it must not masquerade as Dawn-over-other-bands evidence. If no unique fact exists, frame a supported fit-based choice rather than sole logical option. Acceptance: reader can state why Ben preferred this offered device without claiming every other device lacks those controls.

### S04 — preserve handoff, demonstrated setting ability and parent-labor concern

- **Before:** Leah sees she automatically takes over; Evan knows settings; sound still fails to start him (L85–117).
- **Action/location/query:** stool conversation changes search to alarms set without phone (L119–133).
- **Source:** ordinary clock descriptions → “That was how wrist alarms entered the conversation” → “a band made for waking” (L135–153). The connecting source is missing.
- **New fact:** physical cue, strengthening/dismissal, strap/display, no-app setup and motor comparison (L155–197).
- **Objection:** has parental work actually reduced, or moved to setup, charging and enforcing wear?
- **Evidence answer:** no app removes one dependency, not maintenance. “There would still be charging and wearing to remember” (L211) acknowledges the objection but doesn't answer how burdensome either is. Looking at an adjustable strap cannot prove comfort or adherence.
- **Selection:** wrist/strength/controls summarized before “It's called Dawn” (L199–205).
- **P1 contract:** identify actual setup information and specify only verified setup steps relevant to child control. If maintenance evidence is absent, flag it for research, don't invent battery intervals or effortless habits. Keep child's preference distinct from comfort after use. Acceptance: the copy demonstrates or accurately describes why setup needn't require the parent's phone, and does not equate no-app with no work.

### S05 — preserve spare key, June boundary, shared-home search and sauce-call ending

- **Before:** Teresa imagines remote calls or a roommate can preserve home backup; Lena wants an actual separate home (L88–128).
- **Action/location/query:** apartment-floor conversation, lamp unpacking, “quiet alarms for a shared apartment” (L132–138).
- **Source:** search categories then “band details” and “setup instructions” (L138–168). Best contextual route; still no identified result or brand encounter.
- **New fact:** wearer-local cue, gradually increasing vibration, self-set controls, wrist form travels with wearer (L146–176).
- **Objection:** will another person inherit waking; why wrist rather than bed; what if she ignores/removes it away from home; why Dawn?
- **Evidence answer:** roommate labor rightly rejected; wearable preference answers form choice, not superiority; dismissibility admitted. Pre-move practice is sensible preparation, not evidence it will keep working after departure.
- **Selection:** “That's why we chose Dawn Band” (L196), justified by properties of a class of alarms.
- **P1 contract:** add only the source-to-specific-device bridge and supported shortlist rationale. Preserve bed option as legitimate. Do not invent a unique portability moat; many wearables travel. Use existing moving runway for why-now and the guarantee only for financial uncertainty. Acceptance: Dawn is not the first named result after an otherwise complete category choice; no durability conclusion is smuggled through the imagined move.

### S06 — preserve missed call, finite attention and counter realization

- **Before:** Joanne's search implicitly optimizes what she can hear at work, preserving her relay role (L143–161).
- **Action/location/query:** kitchen counter vibration prompts later search for physical rather than sound cues (L153–165).
- **Source:** pillow/wrist options → “When I showed Owen the wrist option” (L181), no named description leading to her confident escalating-vibration answer.
- **New fact:** local physical cue, increasing vibration, on-band settings, maker motor comparison (L173–215).
- **Objection:** will he receive it without her attention; does it require her phone; can she depend on it during a call?
- **Evidence answer:** controls and timed delivery remove particular dependencies; they do not establish waking response. Conditional “if the cue worked for him” (L201) appropriately limits the claim. Backup learning addresses first-use risk, not established independence.
- **Selection:** “The band is Dawn” (L219) after generic-to-particular drift.
- **P1 contract:** trace the actual page/source that supplied the operation description; show which fact narrows the choice beyond a generic wearable. Keep scheduling accommodations legitimate and counter vibration only an observation while awake. Acceptance: no implication that a phone vibrating on a counter validates a sleeping person's response; no reopening of the now-fixed work-schedule continuity issue.

## 5. Full-arc priorities beyond discovery

### P0 — immutable truth and evidence restrictions

1. **No authenticated outcomes or experts in supplied evidence.** Do not invent clinical scenes, customer results, durations to improvement, outside observers of product success or medical explanations to force literal SOP compliance. All six independent-validation cells are unsupported, not invitations to fabricate.
2. **Hardware conflict is unresolved:** each script's notes reports one-minute operation versus approved 'until dismissed' language. The brief authorizes wording but does not resolve physical behavior. Audit cannot silently certify it. Obtain actual current-unit evidence before release; do not explain away the complaint or equate motor strength with three-times waking success.
3. **No exclusive efficacy proof.** No 'only solution,' mandatory getting-up, anti-habituation, automatic independence, ADHD treatment or categorical bed-shaker dismissal.
4. **No invented website contents.** A repair contract may demand an identifiable verified page; it may not dictate that an existing page conveniently contains the needed answer. Missing evidence remains missing until researched under authorization.
5. **Only stated 60-night guarantee.** No refund procedure, free return shipping, no-questions terms, risk-free claim or invented stock pressure. No native prices per user.

### P1 — avoidable craft failures that truth restrictions do not excuse

- **Source continuity:** every band-specific claim needs a credible encounter already staged, not a final URL as retroactive justification.
- **Objection-answer fit:** answer the objection actually opened. More motor force isn't consistency evidence; no app isn't no maintenance; portability isn't unique to Dawn; setting isn't waking; dismissal isn't rising. Acknowledging a concern is not resolving it.
- **Separate category choice from brand choice:** establish physical delivery need, form preference, then particular device fit. Fair alternatives may remain viable. Neither SOP nor Alevia justifies inventing impossible uniqueness.
- **Re-audit rather than merely list:** current sound/preparation re-audits mostly work. Retain them. The next needed depth is relevant purchase alternatives and the new objections created by selecting a wearable, not another long list of failed alarms.
- **Expectations into payoff:** present progression already has practical/emotional stages. Keep conditional entry and callbacks. Add no invented clinical week schedule. Evidence absence is immutable; repetitive 'as mornings add up' can still be improved as craft if a small hypothetical transition is needed, without claiming results.
- **Economics without price:** existing interrupted work, repeated checking, relationship erosion and travel plans supply the cost. A short decision beat can weigh those against learning/wearing/maintenance effort. Do not fabricate wages, purchase totals or promise a refund restores lost time.
- **Risk reversal:** S02 best integrates guarantee with the central hesitation; the others mostly append it. Make it answer the remaining financial-risk question within the narrative, not a substitute for product evidence.
- **Urgency/two roads/PS:** S03's next argument, S05's moving runway and S06's recurring call already give truthful reasons to look now. S01/S02/S04 are softer. Develop the cost of continuing the already-shown routine, not stock scarcity. One genuine PS relationship resolution is enough; no extra PSs solely to match the teardown.

### P2 — preserve variety, don't rename the same scene

S01/S02/S04 converge on child asks → joint search → tapping comparison → dismissal exchange → strap → controls/strength. S01/S03/S04 repeat awake loud-clock demonstration followed by mother hearing it while child sleeps. They are not disqualified for sharing product facts, but repeating the same discovery decisions makes the batch feel authored from a template.

Preserve S01 cupboard reclassification, S02 uncertainty reconstruction, S03 child desk invitation, S04 task ownership, S05 roommate boundary and S06 availability reframe. Differentiate the actual evidence encounter and the decision it changes—not simply the domestic prop or query wording. Do not replace six introductions or six endings to fix two missing middle transitions.

## 6. Surgical release acceptance contract

For each repaired story, a fresh reviewer must be able to answer, using passages rather than notes:

1. What did this mother and child believe before the incident, and what specifically became untenable?
2. What did they do next, where, and with what question?
3. What actual identifiable source did they encounter? What precisely does that source support, and what does it not support?
4. Which new fact changed their understanding of a prior morning or failed attempt?
5. Which objection did it trigger, and where is that objection answered with appropriate evidence or bounded uncertainty?
6. What criteria predate the particular product? Which relevant alternative remains legitimate, and why did this family prefer this device without pretending monopoly?
7. Where exactly is Dawn encountered, rather than merely named? Is each product-specific description connected to that encounter?
8. Do expectations distinguish feeling a cue, dismissing it, rising, remembering to wear it and durable independence?
9. Does the imagined progression resolve the specific incident and relationship without becoming an achieved testimonial? Independent validation remains unavailable unless genuinely sourced separately.
10. Does the guarantee answer only supported financial risk; does urgency come from the actual situation; does the ending preserve broad independence rather than college/age eligibility?

A repair meeting these tests can be called a **better adapted narrative with explicit evidence limitations**, not a literal original-SOP PASS. No automatic gate result, consumer-release clearance or performance forecast is implied.

## 7. Artifacts and verification

Only audit artifacts created:

- `/Users/ayden/.openclaw/workspace/output/fable-six-stories-2026-09-10/alevia-audit/audit.md`
- `/Users/ayden/.openclaw/workspace/output/fable-six-stories-2026-09-10/alevia-audit/matrix.json`

Matrix records current source SHA-256 fingerprints, exact extracted passages, 23-function coverage, source IDs/openings/body ranges and scope boundaries. Primary-word measurements use an explicit regex including URL tokens and are descriptive, not reveal acceptance. No copy, images, CF/site, Meta, skills, memory or canon changes. No paid services.