LLL · task contract

Verification and Constraints

● PARTIAL · wrap-up recorded

Komplex Logopédia: a complete, coherent client-review package so Matt can select the strongest versions and send Rita one clear email.

Review owner: Matt · next review Run: KL-REVIEW-20260924 Updated: 2026-09-25 15:04 CEST

Goal and finish line

End goal: Create the client-facing materials discussed with Rita and Matt, with meaningful A/B choices, one website-derived design system, verified review URLs and an unsent email explaining what Rita should inspect.

Done means: A small review index connects complete ads, landing pages, the email journey, all fourteen daily materials and the testimonial kit. Every required version is readable, linked and tested. Matt has recommended choices and a concise guide. Rita can review the work without mistaking a demonstration or proposal for a live service.

0 / 28criteria verified

7 packagesdeliverables

3 client termsopen decisions

Rules that stay fixed

Matt later wrote “go no”, interpreted as approval to continue. That turn ended during preflight. The subsequent /wrap-up request closes this session with the client package PARTIAL. A future execution must start from preserved work, not claim it already resumed.
Preserve the clickable-demo scope, broad first-pilot audience and Rita’s source constraints. No logged-in platform, invented clinical outcomes, recurring Rita feedback or unsupported testimonials.
No client messages, real uploads, payments, ad activation or binding commitments. Shared paid-API spend stays within USD 15. Preserve unrelated work, private source material and existing review comments.

Proof standard: A checked criterion needs actual files, complete copy, source mapping, working links or inspected screenshots. Local drafts, worker reports and a successful deployment command alone do not prove acceptance. All 28 portfolio checks remain unaccepted.

Deliverable flow

Inputs → outputs · maximum eight nodes

Read left to right. Two source inputs feed the client-review outputs. The session is wrapped up at Matt’s request. Remaining work is recorded in the saved project wrap-up.

Input 1 · confirmed or open

Matt’s settled decisions

Why: Keep the two entry routes, full A/B alternatives and latest scope corrections intact.

Input 2 · constraint or source

Rita’s sources + website

Why: Ground professional content, wording and visual identity in the eight completed audits and original material.

Deliverable 1 · main review item

Ads + landing journeys

Why: Let Rita experience both ways into the offer and Matt compare the complete A/B approaches.

Deliverable 2 · evidence or artifact

Email journey

Why: Show useful content and each transition from video feedback to enrollment.

Deliverable 3 · course

14-day material library

Why: Make every daily unit and original supporting document inspectable.

Deliverable 4 · testimonials

Testimonial kit

Why: Prepare six invitations and a practical interview and consent guide.

Deliverable 5 · review

Verified review index

Why: Gather useful variants, recommendations and tested review URLs in one place.

Output · decision or next action

Matt selects → Rita reviews

Why: The unsent client email explains every URL and the decisions Rita needs to make.

Next: Read the partial wrap-up, then resume from the preserved files when requested. Client email sending remains separate.

Parent ↔ child messages

Newest message first

PARENT · review gate2026-09-24 08:44 CEST

Matt requested the exact template and a review pause. A later “go no” was interpreted as approval to continue, but that turn ended in preflight. His subsequent /wrap-up request closes the session with status PARTIAL.

CHILD · pause confirmed24 September 2026

The collector interrupted the course builder. Acquisition/email and testimonial workers had already finished. No CLI QA agents or publication are in flight. Existing drafts remain preserved. The only published run page is this master board.

Parent read setSmall files first

This contract carries the executable scope. The current partial status and exact recovery steps are in the saved wrap-up report. Supporting source and evidence files stay private in the project. After approval, the executor reads the settled decisions, source inventory and original material before modifying drafts.

Source hierarchy: Matt’s latest corrections → Rita’s direct comments and meeting decisions → original course material → older drafts.

Future completion review requires a clean session gist, deliverable list and main-prompt list. None of these is a substitute for inspecting the actual assets.

Ideas to considerOptional

Later acquisition test

After the first pilot, compare advertising both routes with advertising only the video/list-building entry and selling cohorts by email.

Confirm operating capacity

Propose a practical free-video intake cap. Rita confirms workload before any three-working-day promise goes live.

Keep review simple

Recommend the strongest variants inside the shared index. Leave alternatives accessible without overwhelming the client email.

Work and proof

One card per deliverable

Create a review index for selection

Give Matt and Rita one simple route through the package, with meaningful alternatives and a single visual identity.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Provide a complete client index and a separate private morning overview with recommended defaults and reasons.
Derive shared typography, colors, logo, components and imagery from the current website. Apply them consistently across every new asset.
Present two meaningfully different versions of every substantive new asset. Existing original documents may be shared rather than duplicated.
Open the final review pages in the in-app browser and keep client-visible content free of raw transcripts, internal prompts and private data.

Protect

× One coherent brand system, not unrelated A/B designs.
× Use a small useful set of review URLs. Keep private operational notes out of public asset roots.

Evidence: Evidence required: version inventory, shared-style source map, inspected index, actual browser-open receipt and public-file privacy check. Local index exists but is not accepted.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Build acquisition journeys to explain both offers

Deliver full free-video and direct-challenge paths from ad to simulated enrollment.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Create at least two complete ad texts and two usable creative assets for each entry route, with matching landing messages.
Create two complete free-video landings and two challenge landings, with genuinely distinct copy or structure.
Make both journeys clickable through demo receipt, feedback example, challenge, checkout and enrollment confirmation, including error, full and expired states.
Show separate marketing opt-in, precise service conditions and clearly labelled proposed terms. Test that no real upload, purchase or enrollment occurs.

Protect

× Ads say short video and personal feedback. Do not lead with unnecessary names, time limits or written-format filler.
× 24,900 Ft, 12 places and three working days are Matt’s proposals for Rita. No confirmed date or visitor-reset scarcity. No logged-in platform.

Evidence: Evidence required: full A/B copy, creative files, link/state test results, side-effect check and desktop/mobile deployed screenshots. Local acquisition package exists, not accepted.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Write emails to guide the next step

Make the complete communication sequence useful, natural and ready for Rita’s content review.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Write two complete variants at every stage: receipt, example personal feedback, content-led nurture, weekly newsletter, cohort promotion, enrollment and next steps.
Build each helpful nurture message around a practical speech-related observation before connecting naturally to the challenge.
Record triggers, timing, opt-in and buyer suppression privately. Cohort promotion replaces the weekly newsletter. Include source-supported closing-consultation follow-up.
Independently review every full variant in journey order for Hungarian fluency, repetition, offer consistency, valid links and correct transitions.

Protect

× No fabricated personal assessment. Feedback remains Rita’s work, with examples explicitly labelled.
× No forced sentimentality, hype, em dashes or semicolons. No actual sending or invented deadline.

Evidence: Evidence required: stage-by-variant matrix, complete Markdown drafts, designed previews, trigger map and independent copy review. Local email package exists, not accepted.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Assemble daily materials to support practice

Preserve Rita’s six modules in fourteen daily units and a complete downloadable review library.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Provide fourteen coherent daily units covering breathing, articulation, voice/resonance, rhythm, speech and movement, and integration.
Preserve all relevant original professional content, exercises and cautions, with a source map and working source-document downloads.
Provide two complete editorial presentations of the same canonical material, daily reminder copy alternatives and closing-consultation guidance.
Clearly separate optional baseline, participant-held practice recordings and at most one defined 2–3 minute closing clip. Mark exact daily ordering and delivery mechanics as review proposals.

Protect

× Do not replace long-form source content with generic short summaries or invent clinical exercises to create variants.
× No daily Rita feedback, clinical guarantees or claimed live Messenger/login integration. No unsupported daily time commitment.

Evidence: Evidence required: fourteen-day/module/source matrix, content comparisons, document/download checks and complete deployed section screenshots. Course artifacts were interrupted and need inspection before reuse.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Prepare testimonials to collect useful feedback

Give Rita a complete invitation and interview kit and account for earlier client promises.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Provide all six complete invitation variants and two coherent kit presentations using the shared design.
Include neutral interview questions and clearly separate participation, recording consent and later public-use consent.
Use the verified service-specific booking URL in invitations. Verify read-only without booking or changing the service.
Account for existing reporting and therapy-booking promises with verified relevant links or precise gaps in a concise secondary review section.

Protect

× No invented testimonials, performance report data or unverified appointment duration in invitation copy.
× Reuse finished relevant work. Do not change live booking or create a new reporting project merely to fill the package.

Evidence: Evidence required: six-invitation inventory, consent/interview text, verified Reservio page and prior-delivery source receipts. Local kit and booking readback exist, final acceptance pending.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Verify hosted pages to prove readiness

Prove that the assembled package works and reads correctly on the actual review host.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Run independent Hungarian copy and source-fidelity review across the full customer journey. Resolve material contradictions.
Check every navigation path, A/B control, demo/error/success state and download, including keyboard focus and prohibited side effects.
Use at least two independent actual gpt-5.6-luna xhigh reviewers across every page, visible section, tall segment, desktop/mobile viewport and meaningful state. Save and inspect section PNGs.
Fix defects with actual gpt-5.6-luna max and fresh QA. Verify actual postdeployment screenshots, stable URLs, review toolbar/comment persistence and an unaffected existing host page.

Protect

× Do not substitute another model and call it Luna. Preserve evidence of unverified checks.
× One publication owner preserves the shared host and stable comment identity. No force deployment that drops unrelated pages.

Evidence: Evidence required: complete PASS/FAIL/UNVERIFIED coverage tables, model receipts, local and deployed PNGs, fix/recheck records and disposable comment/rewrite persistence proof. Not yet complete.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Draft Rita’s email to explain each link

Make the morning selection and client review easy without sending anything prematurely.

PAUSED0 / 4

Show criteria, constraints, and evidence

Verify

Provide two complete unsent Hungarian email drafts to Rita with real verified review URLs and precise instructions for each.
Put the recommended draft and default selections first, explaining the main principles and exact questions Rita should answer.
Provide one private overview with asset inventory, variant recommendations, verification results, remaining proposals and a resumable executor handoff.
After the package is verified, refine the Markdown narration and produce one local Charon guide at 1.25× after two sequential independent Luna Max reviews.

Protect

× Do not claim drafts are an activated campaign or that Matt’s proposals have Rita’s approval.
× No automatic email, Telegram or other external sends. One combined final guide, no per-child audio.

Evidence: Evidence required: unsent email files with final links, easy review order, accurate status ledger, reviewed transcript and decoded audio. Local email drafts need final URLs and verification before acceptance.

Screenshot evidence: Pending where visual proof is required. No deployed QA pass is claimed.

Open decisions and blockers

Only unresolved items

CURRENT STATE · partial wrap-upExecution paused

The contract remains the review record. Local client drafts exist, but none of the five planned client pages has been published or passed deployed visual QA. The exact recovery route is in the saved project wrap-up.

PROPOSALS FOR RITA · no execution gateClearly labelled in client review

24,900 Ft pilot price, 12 places, short written free-video feedback within three working days, the precise daily ordering and intended Messenger delivery remain proposals. The actual start/closing dates, intake cap and final consultation details are open. A future executor may recommend options without inventing Rita’s acceptance.

Recovery state

Only inspect on stall or completion

0consecutive stalls

PAUSEDrecovery state

Preserved draftsresume point

Last stall / completion note: Matt’s later go-on message did not lead to an executed build before the turn was interrupted. The wrap-up records the local drafts and missing publication/QA. No finished client package is claimed.

Parent action: Keep the one-shot scheduler paused. On a future continuation, reuse the preserved work and reconcile the interrupted course owner. Do not duplicate the old build.