# From an entity-source recording to a client reading script

Use when Margarita provides a long recording about entities / subtle presences, asks to isolate one type, and then asks for a text she can read aloud to clients.

## Fast source workflow

1. Download and probe the whole recording.
2. For urgent targeted extraction, split it into 5–10 minute chunks and run a fast first-pass transcription across all chunks.
3. Search inflected and ASR-distorted variants of the requested term. For `деградирующие сущности`, also search roots and semantic neighbours: `деград`, `разлаг`, `разруш`, `дряхл`, `старен`, `распад`, `сущност`.
4. Convert chunk-local timestamps to global recording timestamps before reporting them.
5. Re-transcribe only the relevant chunks with a more accurate model and compare against the fast pass before treating specialist wording as source-exact.
6. Distinguish clearly between:
   - what the source explicitly says;
   - cleaned reconstruction where ASR was uncertain;
   - the new Cosmos Luminosity client script.

## Client-script architecture

When Margarita says `дай текст, который я буду читать клиентам`, return a continuous, copy-ready spoken script rather than another summary.

Recommended arc:

1. brief body and present-time orientation;
2. transparent framing of the energetic/symbolic model;
3. permission not to visualize or believe anything;
4. differentiation: physical need, emotion, memory, body response, residual energetic trace, subtle presence, or another configuration;
5. identify function and whether it is complete;
6. preserve the ecological role of decomposition and completion;
7. release only what is currently complete and no longer fitting;
8. close old access routes and restore current consent;
9. return attention/resources to body and space;
10. ask for one concrete earthly care step;
11. full reorientation to room/body/present time;
12. short scope boundary: symbolic practice does not replace medical or psychotherapeutic care.

## Language and ethics

- Keep the requested term `деградирующие сущности` when it is the source category, but explain functionally and without demonization.
- Do not call them evil, parasites, enemies, or universally harmful.
- Preserve the source distinction: decomposition can be ecologically appropriate; intervention concerns presence/function in a place where it is no longer required or disrupts the whole system.
- Do not instruct the facilitator to remove something from every client automatically. Use current consent, body response, and appropriateness.
- Do not frame illness, aging, pain, fatigue, or psychiatric symptoms as proof of an entity. Explicitly differentiate ordinary physical and psychological explanations.
- Do not promise rejuvenation, healing, slowed aging, repaired buildings, or ecological restoration.
- Keep release and collaboration separate. A script about completed destructive/decomposition functions should not silently invite future cooperation.
- Use long, calm spoken sentences and natural pauses; avoid a choppy list of mystical commands.
- If exact clearing formulas are being canonized for the manual, return to phrase-by-phrase approval. A one-off client reading script may be drafted as a whole, but it remains a draft until Margarita approves it.

## Delivery rule

If the user asks `для копирования`, provide only the script, with no prefatory analysis or postscript. If she asks for both explanation and script, put the explanation first and the uninterrupted copy block last.