Show Notes From the Transcript You Pasted

<context> I need a three-sentence episode summary for show notes. Claims only, from the transcript. This is not titles, not chapters, and not a newsletter issue. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Episode number if I have it: [EPISODE NUMBER OR NONE] - Guest label: [GUEST LABEL OR SOLO] - House tone: [PLAIN / WARM / DRY] - Words I never use: [BANNED WORDS OR NONE] </inputs> <task> Write a three-sentence summary of [PASTE TRANSCRIPT]. Sentence 1: who this is for and the problem as spoken. Sentence 2: the main claim as spoken. Sentence 3: what a listener can do with it. Then write a tighter two-sentence alt. Mark any sentence that is a topic label rather than a claim and rewrite it. If a claim is not in the paste, drop it. </task> <constraints> - Do not invent a quote, number, guest, or result. - Do not write we discussed X, we cover Y, or in this episode. - Honor [BANNED WORDS]. - No timestamps unless I pasted them and you are citing a source line, not inventing a clock. </constraints> <format> Three-sentence summary, two-sentence alt, then a source tag per sentence. </format>

<context> Apple and many apps truncate show-note descriptions after about two lines. I need those two lines first. Summary copy only. Not a title slate and not chapters. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Guest label: [GUEST LABEL OR SOLO] - Phrase I can defend in the hook: [PHRASE OR NONE] - Audience: [AUDIENCE] - Character target for the two lines: [UNDER 220] </inputs> <task> Write 6 two-line hooks from [PASTE TRANSCRIPT]. Each pair must stand alone when truncated. Put a spoken claim or spoken problem in line 1. Use [PHRASE] only if it is in the paste or I supplied it. Count characters for the pair. Rank a top 2 I can paste at the top of the host description. </task> <constraints> - Do not invent a quote or number to win the hook. - Do not start with Welcome back or In this episode. - Do not write the rest of the description here. - If the tape has no stand-alone claim, say so and write a problem hook from spoken pain only. </constraints> <format> Numbered hooks with character counts and source tags, then the two to paste into the host. </format>

<context> I need show-note takeaways written as claims the speakers made. Topic labels are a fail. This is not a title list and not a newsletter. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - How many takeaways I want: [COUNT] - Guest label: [GUEST LABEL OR SOLO] - Audience: [AUDIENCE] - Claims I wish I could make: [WISH LIST OR NONE] </inputs> <task> Write up to [COUNT] takeaways from [PASTE TRANSCRIPT]. Each takeaway is one sentence that states the claim, not the theme. Tag each to a source line. If [WISH LIST] items are not in the tape, list them as UNSUPPORTED and do not publish them. If honest claims are fewer than [COUNT], stop. Do not fill the gap. </task> <constraints> - Do not invent a claim, quote, or number. - Reject any line that only names a topic (pricing, culture, AI). - Do not add a timestamp you did not paste. - No CTA paragraph in this pass. </constraints> <format> Numbered claim takeaways with source tags, then UNSUPPORTED wishes, then a shortfall note if [COUNT] cannot be met. </format>

<context> I need a listen-if and skip-if blurb so a busy person can decide in ten seconds. Base it on who the transcript actually serves. This is summary copy, not a promo calendar. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Who I think it is for: [AUDIENCE] - Who I think should skip: [SKIP AUDIENCE OR UNKNOWN] - Guest label: [GUEST LABEL OR SOLO] - Length: [UNDER 80 WORDS] </inputs> <task> Write a listen-if paragraph and a skip-if paragraph from [PASTE TRANSCRIPT]. Listen-if must name a person and a job the episode actually helps. Skip-if must name a person the tape does not serve. If [SKIP AUDIENCE] is UNKNOWN, infer only from what the tape never covers, and mark INFERRED. Then write a 25-word combined blurb for the top of the host description. </task> <constraints> - Do not invent a listener persona the tape does not support. - Do not invent a quote or timestamp. - Do not claim the episode covers a topic that never appears. - No subscribe-now closer in this pass. </constraints> <format> Listen-if, skip-if, 25-word blurb, then INFERRED flags. </format>

<context> I paste one blob into every field today. I need four self-contained blurbs from the same transcript for different host and site fields. Summary only. No chapter list. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Guest credit line I already approved: [GUEST CREDIT OR NONE] - Phrase I want in the 150-word version if spoken: [PHRASE OR NONE] - Call to action I am allowed to use: [CTA OR NONE] - Audience: [AUDIENCE] </inputs> <task> Write four blurbs from [PASTE TRANSCRIPT]: 15 words, 50 words, 150 words, 400 words. No sentence reused across versions. The first 120 characters of the 50, 150, and 400 must stand alone when truncated. Use [GUEST CREDIT] only as I wrote it. Use [PHRASE] only if it is in the paste or I supplied it. Put [CTA] only on the 150 and 400, late, and only if I gave one. </task> <constraints> - Do not invent a quote, number, link, or timestamp. - Do not write In this episode we discuss. - Count words on each version and stop at the cap. - If [CTA] overclaims the tape, weaken it and say why. </constraints> <format> Four labeled blurbs with word counts, then a source note for the lead claim in each. </format>

<context> I need chapter markers for show notes. Use only timestamps that already appear in the pasted transcript. If there are none, stop. This is not titles and not a quote sheet. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Episode length if I have it: [DURATION OR UNKNOWN] - House chapter style: [SHORT LABELS / DESCRIPTIVE] - Platforms I will paste into: [APPLE / SPOTIFY / YOUTUBE / HOST] </inputs> <task> If [PASTE TRANSCRIPT] has no mm:ss or h:mm:ss stamps, return NO TIMESTAMPS IN PASTE and stop. Do not invent 0:00 or any other clock. If stamps exist, build 8 to 14 chapters. Each row: start time copied exactly from the paste, a label under 45 characters that names the payoff spoken there, and the source line. Break at topic changes, not at round minutes. </task> <constraints> - Do not invent, round, or interpolate a timestamp. - Do not move a stamp to a neater clock (4:00 instead of 4:07). - Do not write quotes in this pass. - If stamps look wrong versus [DURATION], flag UNRELIABLE and still do not guess a replacement. </constraints> <format> Chapter table (start, label, source) or NO TIMESTAMPS IN PASTE. Then UNRELIABLE flags. </format>

<context> I already have chapter start times in the transcript. The labels are generic. Rewrite labels as payoffs. Do not change or invent times. This is timestamps only. </context> <inputs> - Transcript with existing stamps: [PASTE TRANSCRIPT] - Current labels if I have them: [CURRENT LABELS OR NONE] - Episode promise in my words: [PROMISE] - Style: [SHORT LABELS / DESCRIPTIVE] </inputs> <task> If there are no timestamps in [PASTE TRANSCRIPT], return NO TIMESTAMPS IN PASTE and stop. Otherwise keep every start time exactly as pasted. Rewrite each label so it names the payoff spoken in that span, not the topic. Labels under 45 characters. If a span has no spoken payoff, keep a dull honest label rather than inventing a hook. </task> <constraints> - Do not invent a timestamp. - Do not invent a quote to put in the label. - Do not merge or split chapters by guessing new clocks. - Do not write a full summary. </constraints> <format> Before/after table (time unchanged, old label, new label, source line) or NO TIMESTAMPS IN PASTE. </format>

<context> I have a timestamped transcript. I need the same chapters formatted for Apple, Spotify, YouTube, and a generic host field. Times must stay as pasted. This is not new chaptering from a blank tape. </context> <inputs> - Transcript with stamps: [PASTE TRANSCRIPT] - Chapter labels I already approved, if any: [LABELS OR NONE] - Platforms: [APPLE / SPOTIFY / YOUTUBE / HOST] - Episode length: [DURATION OR UNKNOWN] </inputs> <task> If [PASTE TRANSCRIPT] has no timestamps, return NO TIMESTAMPS IN PASTE and stop. If it has them, output the same start times in each platform's common paste shape (0:00 Title lines for YouTube and many hosts; a simple start plus label list for Apple/Spotify notes). Use [LABELS] if I pasted them; otherwise write payoff labels from spoken content. Do not add a chapter whose start time is not in the paste. </task> <constraints> - Do not invent, convert, or beautify a time. - Do not add 0:00 unless 0:00 (or 00:00) is in the paste. - Do not write descriptions or quotes. - If [DURATION] conflicts with a stamp, flag it. Do not fix the clock. </constraints> <format> One labeled paste block per platform, then a shared warning list. </format>

<context> I need sponsor, mid-roll, and CTA chapter markers. Only mark a break if the transcript already has a timestamp at that moment and the spoken line is actually a break. Do not invent a mid-roll clock. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Breaks I know happened: [BREAK LIST OR UNKNOWN] - Sponsor name if said on tape: [SPONSOR OR NONE] - CTA I used on tape: [CTA OR NONE] </inputs> <task> If there are no timestamps in [PASTE TRANSCRIPT], return NO TIMESTAMPS IN PASTE and stop. Scan for spoken sponsor reads, mid-roll handoffs, and closing CTAs. For each, copy the nearest existing timestamp from the paste (do not interpolate). Label it as Sponsor, Mid-roll, or CTA. If [BREAK LIST] names a break with no matching stamp or spoken line, mark CANNOT MARK. </task> <constraints> - Do not invent a timestamp for a break I remember but did not stamp. - Do not invent a sponsor name that is not in the paste or in [SPONSOR]. - Do not write ad copy. - If the CTA was not spoken, do not add a CTA chapter. </constraints> <format> Break table (time from paste, label, spoken evidence) then CANNOT MARK rows. </format>

<context> A platform or host field wants fewer chapters. I need a six-chapter cut that only uses start times already in the transcript. This is a subset, not a rewrite of the clocks. </context> <inputs> - Transcript with stamps: [PASTE TRANSCRIPT] - Full chapter list if I have one: [FULL LIST OR NONE] - Episode promise: [PROMISE] - Platforms with a chapter cap: [PLATFORM] </inputs> <task> If [PASTE TRANSCRIPT] has no timestamps, return NO TIMESTAMPS IN PASTE and stop. Choose exactly 6 start times from the stamps that already exist (or from [FULL LIST] if those times also appear in the paste). Prefer topic changes that carry [PROMISE]. Write a payoff label under 45 characters for each. List the stamps you dropped and why. Do not create a new time to make the six look even. </task> <constraints> - Do not invent or average timestamps. - Do not keep a chapter whose start is not in the paste. - Do not add quotes. - If fewer than 6 honest stamps exist, output fewer and say why. </constraints> <format> Six-row paste list, dropped stamps with reasons, or NO TIMESTAMPS IN PASTE. </format>

<context> I need pull-quotes for show notes. Only words that appear in the transcript may be quoted. This is not a title pass and not a paraphrase exercise. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - How many quotes I want: [COUNT] - Speakers I already labeled: [SPEAKER LIST] - Theme if I have one: [THEME OR NONE] - Quotes I already approved: [APPROVED QUOTES OR NONE] </inputs> <task> Extract up to [COUNT] verbatim quotes from [PASTE TRANSCRIPT]. For each: exact wording, speaker from [SPEAKER LIST] only, a one-line context, and a timestamp only if that stamp is already next to the line in the paste. If [THEME] is not in the tape, say so and pull the strongest honest lines instead. If fewer than [COUNT] honest quotes exist, stop. Do not fill the gap. </task> <constraints> - Do not invent, tidy, complete, or punch-up a quote. - Do not assign a speaker who is not in the paste or in [SPEAKER LIST]. - Do not invent a timestamp. - Shorten only with an ellipsis and show the cut. </constraints> <format> Quote cards (wording, speaker, context, timestamp or NO STAMP IN PASTE), then a shortfall note. </format>

<context> I will put quotes in show notes and a newsletter. Each quote needs the minimum context so it does not invert the speaker. Use only the paste. Do not invent clocks. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Speakers I labeled: [SPEAKER LIST] - Where I will use them: [HOST / NEWSLETTER / BOTH] - Sensitive topics to avoid: [OFF LIMITS OR NONE] </inputs> <task> Pull 8 verbatim quotes from [PASTE TRANSCRIPT]. For each: exact words, speaker from [SPEAKER LIST], minimum context (one sentence, paraphrase marked PARAPHRASE), timestamp only if present in the paste, and a use tag (host notes, newsletter, both, or DO NOT USE). Honor [OFF LIMITS]. Rank the 3 that can stand alone. </task> <constraints> - Do not invent a quote or a timestamp. - Do not write a sharper version of the line. - If context is missing from the paste, mark CONTEXT UNKNOWN. Do not invent the setup. - Do not write a full newsletter issue here. </constraints> <format> Eight cards, then top 3, then DO NOT USE rows. </format>

<context> I need quotes grouped by who said them. If a speaker is not labeled in the paste or in my list, do not guess. This is a quote inventory for show notes, not a script. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Host label: [HOST LABEL] - Guest label: [GUEST LABEL OR SOLO] - Other speakers I already named: [OTHER SPEAKERS OR NONE] - How many per person: [COUNT] </inputs> <task> Split verbatim quotes from [PASTE TRANSCRIPT] into Host, Guest, Other, and SPEAKER UNKNOWN. Take up to [COUNT] per bucket. Timestamp a quote only if the stamp is in the paste. If SOLO, put guest bucket as EMPTY, SOLO EPISODE. Do not promote an unknown line into [HOST LABEL] or [GUEST LABEL]. </task> <constraints> - Do not invent a speaker, quote, or timestamp. - Do not clean a line to sound more like the host brand. - If labels conflict, show both and stop. Do not pick a winner. - No newsletter copy in this pass. </constraints> <format> Four buckets of quote cards, then a list of lines that must not run until labeled. </format>

<context> I need quote-card lines for a newsletter or show-notes image. Trims must stay true. This is not design and not a new quote. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Speakers: [SPEAKER LIST] - Target words per card: [UNDER 22] - Attribution line I already use: [ATTRIBUTION] - Quotes I must not use: [OFF LIMITS OR NONE] </inputs> <task> From [PASTE TRANSCRIPT], pick up to 6 lines that can survive as cards. For each: verbatim full line, trimmed card line at [UNDER 22] words, a mark on every omitted word, attribution from [ATTRIBUTION], and a timestamp only if pasted. Reject any line that needs hidden context to stay true. Honor [OFF LIMITS]. </task> <constraints> - Do not invent a quote. - Do not change meaning to hit the word cap. If it will not trim honestly, reject it. - Do not invent a timestamp. - Do not write alt-text that adds a claim. </constraints> <format> Card table (full, trim, cuts marked, attribution, stamp or NO STAMP), then rejected lines. </format>

<context> I need an honesty pass on quotes before they hit show notes or a newsletter. Flag lines that mislead out of context, joke as fact, or need approval. Do not invent safer quotes to replace them. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Speakers: [SPEAKER LIST] - Sensitive topics: [OFF LIMITS OR NONE] - Guest approval status: [APPROVED / NOT ASKED] - Where I planned to use quotes: [HOST / NEWSLETTER / SOCIAL] </inputs> <task> Read [PASTE TRANSCRIPT] and build three lists: CLEAR TO QUOTE (verbatim, speaker known, stands alone), NEEDS CONTEXT (usable only with a pasted setup sentence), and DO NOT QUOTE (misleading, [OFF LIMITS], speaker unknown, or needs approval if [NOT ASKED]). Timestamp only if present. Do not rewrite a DO NOT QUOTE line into a punchier clear quote. </task> <constraints> - Do not invent a quote or timestamp. - Do not tidy a risky line so it becomes clear. - If [NOT ASKED], keep named-guest zingers in NEEDS CONTEXT or DO NOT QUOTE, not CLEAR. - No social captions in this pass. </constraints> <format> Three lists with wording, speaker, stamp or NO STAMP, and reason. </format>

<context> I need the first two lines of the episode description for Apple and Spotify search. They must come from the transcript. This is SEO copy for the host, not a newsletter and not YouTube Studio. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Primary phrase I want: [PRIMARY KEYWORD OR NONE] - Audience: [AUDIENCE] - Guest label: [GUEST LABEL OR SOLO] - Character target: [UNDER 180] </inputs> <task> Write 5 first-two-line options from [PASTE TRANSCRIPT]. Each pair under [UNDER 180] characters, standing alone when truncated. Include [PRIMARY KEYWORD] only if it appears in the paste or I typed it in this input. Name the spoken payoff. Count characters. Pick one paste-ready pair for the top of the host description. </task> <constraints> - Do not invent a keyword, quote, number, or timestamp. - Do not start with In this episode. - Do not write the rest of the SEO body in this pass. - No hashtag dump. </constraints> <format> Five options with counts and source tags, then the pair to paste into the host. </format>

<context> I need a full episode description for the podcast host: hook, body, takeaways, CTA. Every claim must be in the transcript. This is not a blog post and not a YouTube description pack. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Primary phrase: [PRIMARY KEYWORD OR NONE] - Related phrases I already chose: [RELATED PHRASES OR NONE] - Guest credit: [GUEST CREDIT OR NONE] - CTA I am allowed: [CTA OR NONE] - Length: [WORD COUNT] </inputs> <task> Write one paste-ready description of about [WORD COUNT] from [PASTE TRANSCRIPT]: (1) two-line hook, (2) short body that uses [PRIMARY KEYWORD] and [RELATED PHRASES] only if they are in the paste or I supplied them, (3) 5 claim takeaways, (4) [GUEST CREDIT], (5) [CTA] late if I gave one. If a phrase is not in the paste and I did not supply it, skip it. </task> <constraints> - Do not invent a quote, number, link, or timestamp. - Do not stuff the same phrase three times. - Do not write chapter clocks unless those stamps are in the paste and I ask for them in a later prompt. - If [CTA] overclaims the tape, weaken it. </constraints> <format> Paste-ready description, word count, then a phrase-to-source list. </format>

<context> I need episode tags and search phrases that a listener might type, taken from words already spoken. Do not invent a ranking target. This is SEO inventory, not titles. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Phrases I already want: [WISH LIST OR NONE] - Show niche I already use: [NICHE] - How many tags the host allows: [TAG LIMIT] </inputs> <task> From [PASTE TRANSCRIPT], list spoken phrases that could work as directory tags or in-description terms. For each: the exact phrase, a source line, and a note (use in body, use as tag, or too generic). Check [WISH LIST]: KEEP if present in the paste or I supplied it as an input I will stand behind; DROP if it is a new invented topic. Return a tag list that fits [TAG LIMIT]. </task> <constraints> - Do not invent a keyword the speakers never used unless I put it in [WISH LIST] as my own target, and then mark SUPPLIED BY ME, NOT IN TAPE. - Do not invent a quote or timestamp. - Do not write the full description. - Prefer specific spoken phrases over generic podcast tags. </constraints> <format> Spoken-phrase table, wish-list KEEP/DROP, then the paste-ready tag list. </format>

<context> Guests name books, tools, people, and sites. I need a show-notes resource list. Never invent a URL, author, or year. This is SEO and notes hygiene, not a newsletter. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Links I already have: [LINK LIST OR NONE] - Glossary of names I already know: [GLOSSARY OR NONE] - Guest label: [GUEST LABEL OR SOLO] </inputs> <task> Extract every book, tool, paper, company, person, and framework named in [PASTE TRANSCRIPT]. For each: name as spoken, how they framed it, timestamp only if pasted, and URL = use [LINK LIST] if it matches, else BLANK. Mark SPELLING UNCLEAR when the paste is messy. Do not guess a publisher or a .com. </task> <constraints> - Do not invent a URL, author, or publication year. - Do not invent a timestamp. - Do not write your own blurb for a tool. Keep the guest's framing. - If [GLOSSARY] conflicts with the tape, show both and stop. </constraints> <format> Grouped list with BLANK url cells, SPELLING UNCLEAR rows, then a fill-in table for me. </format>

<context> I need episode-level category, subcategory if any, and tags for the podcast host. Recommendations must follow what was actually spoken. This is field copy, not a marketing plan. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Show-level category I already use: [SHOW CATEGORY] - Audience: [AUDIENCE] - Host fields I can fill: [CATEGORY / TAGS / AUTHOR / NONE] </inputs> <task> From [PASTE TRANSCRIPT], recommend: primary episode category (say if it should stay [SHOW CATEGORY]), optional secondary, and up to 10 tags from spoken phrases. For each recommendation, cite a source line. If the tape does not support a category change, say KEEP SHOW CATEGORY. Write the author/guest field only from labels I supplied or names in the paste. </task> <constraints> - Do not invent a category to chase a trend the tape never covered. - Do not invent tags. - Do not invent a quote or timestamp. - Do not write the long description here. </constraints> <format> Field-by-field paste lines (category, tags, author), then source tags, then KEEP/CHANGE note. </format>

<context> I need a newsletter issue cut from a podcast transcript. The reader may never press play. Use only the paste plus my take. This is not a full show-notes pack and not a social thread. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - The one idea I want to expand: [IDEA] - My take: [HOST TAKE] - Audience: [AUDIENCE] - CTA I am allowed: [LISTEN / REPLY / SHARE / NONE] - Guest label: [GUEST LABEL OR SOLO] </inputs> <task> Write a paste-ready newsletter issue: subject plus 3 alternates, preview text, opening, the [IDEA] expanded with [HOST TAKE], one quoted exchange only if both lines appear in the paste, 3 claim takeaways, a listen-if and skip-if line, and [CTA]. If [IDEA] is not in the tape, stop and say UNSUPPORTED IDEA. </task> <constraints> - Do not invent a quote, number, or timestamp. - Subject under 45 characters and never episode is live. - The issue must teach something without the audio. - No second CTA. </constraints> <format> Subject options, preview, full issue, then source tags for the idea and any quote. </format>

<context> I need newsletter subject lines for an episode issue. Each subject must be a claim someone actually said, or a tight paraphrase marked PARAPHRASE. This is subjects only. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - My take if I want it in the subject: [HOST TAKE OR NONE] - Words I will not use: [BANNED WORDS OR NONE] - Audience: [AUDIENCE] - Character comfort: [UNDER 45] </inputs> <task> Pull spoken claims from [PASTE TRANSCRIPT]. Write 10 subject lines under [UNDER 45] characters. Tag each VERBATIM claim, PARAPHRASE, or HOST TAKE. Host-take subjects are allowed only if I pasted [HOST TAKE] and it does not invent a result. Rank a top 3. Write matching preview text (under 90 characters) that does not repeat the subject. </task> <constraints> - Do not invent a hook the tape does not support. - Do not invent a quote or timestamp. - Honor [BANNED WORDS]. - No episode 12 is live subjects. </constraints> <format> Ten subjects with counts and tags, preview lines for the top 3, then the one to paste into the newsletter. </format>

<context> I need one quoted exchange for the newsletter: host line, guest line, maybe a follow-up. Every line must appear in the transcript. Do not invent dialogue. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Speakers: [SPEAKER LIST] - Theme I want the exchange to illustrate: [THEME] - How many turns: [2 TO 4] - Guest approval: [APPROVED / NOT ASKED] </inputs> <task> Find one exchange in [PASTE TRANSCRIPT] on [THEME]. Rebuild [2 TO 4] turns with verbatim lines, speaker labels from [SPEAKER LIST], and timestamps only if pasted. If [THEME] has no honest exchange, say NONE and stop. If [NOT ASKED] and the guest is identifiable, mark NEEDS APPROVAL. Write a one-sentence setup that does not add facts. </task> <constraints> - Do not invent a line, a speaker, or a timestamp. - Do not merge two moments into one cleaner fight. - Do not punch-up wording. - If a turn is missing, leave it out. Do not write the missing line. </constraints> <format> Setup sentence, labeled turns, stamp or NO STAMP each, then NEEDS APPROVAL if required. </format>

<context> Open rates fall when I only announce the episode. I need a cut that teaches one idea from the transcript plus my take, even if nobody clicks play. Newsletter copy only. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Idea to expand: [IDEA] - My take: [HOST TAKE] - Audience: [AUDIENCE] - Word count: [WORD COUNT] - Link to the episode: [LINK OR PLACEHOLDER] </inputs> <task> Write a newsletter body of about [WORD COUNT]: open on the problem as spoken, teach [IDEA] from the tape, add [HOST TAKE] as a separate labeled take (do not put words in the guest's mouth), one verbatim quote only if it exists, three actions a reader can do this week, then a listen-if line with [LINK OR PLACEHOLDER]. If [IDEA] is not in the paste, stop. </task> <constraints> - Do not invent a quote, number, or timestamp. - Do not attribute [HOST TAKE] to the guest. - Do not write a 400-word recap of the whole hour. - If you want a second quote and it is not in the paste, skip it. </constraints> <format> Paste-ready body, word count, source tags, then a note on which paragraph is only my take. </format>

<context> I send two emails from one episode: a short publish-day teaser and a later full cut. Both must come from the same transcript. No new quotes. No invented clocks. </context> <inputs> - Transcript: [PASTE TRANSCRIPT] - Idea: [IDEA] - My take: [HOST TAKE] - Guest label: [GUEST LABEL OR SOLO] - CTA: [LISTEN / REPLY / NONE] </inputs> <task> Write two paste-ready newsletter versions from [PASTE TRANSCRIPT]. Teaser: under 90 words, subject, preview, one spoken claim, [CTA]. Full cut: 250 to 400 words, different subject and different open, expanded [IDEA] plus [HOST TAKE], one verbatim quote only if present, listen-if and skip-if. No reused first sentence. </task> <constraints> - Do not invent a quote, number, or timestamp. - If [IDEA] is missing from the tape, stop. - Do not make the teaser a truncated copy of the full cut. - No extra links I did not supply. </constraints> <format> Teaser block (subject, preview, body), full-cut block (subject, preview, body), then source tags shared by both. </format>