- Home
- Skills
- Translation & Localization
- From Transcript to Deliverable: Summaries, Quotes, and Subtitles
More screenshots
Works with the AI Agents you already use
From Transcript to Deliverable: Summaries, Quotes, and Subtitles
Turn a transcript you already have into the artifact you need: layered summaries, quote packs, subtitle files, translations, fact-check briefs, or chapter maps.
$5
From Transcript to Deliverable: Summaries, Quotes, and Subtitles
Example session with this skill installed
I have the transcript of a fictional 38-minute talk by June Park, a fictional community organizer with a fictional group called Maple Street Mutual Aid, about starting a neighborhood tool library. Give me a chapter map and the three best quotes, with timestamps.
- Read your context and instructions
- Compiled the from transcript to
CHAPTER MAP + QUOTES — June Park, "Starting a Neighborhood Tool Library," 38-minute talk. Invented scenario and numbers for this demo only.
Chapters: Why borrow (00:00–09:00) — the case for sharing tools instead of buying them. The first shelf (09:00–19:00) — 212 tools gathered from 64 households on a $340 first-year budget. Lending rules that held (19:00–29:00) — the three rules that survived the first year. What broke and what changed (29:00–38:00) — 3 broken drills and the repair fund they led to.
Quotes, each verified against the raw text:
- "A drill you use twice a year should not cost you a Saturday's pay." — June Park, 04:18
- "The shelf filled faster than the rules did." — June Park, 13:52
- "We lend the tool, not the lesson — the lesson comes back with it." — June Park, 26:30
First-run faults fixed: the first chapter title was named after a joke in the middle of the chapter and was renamed to "Why borrow"; the first quote draft was not self-contained and was expanded to the full sentence; the second quote's timestamp was first written as 13:58 and corrected to 13:52 on re-check against the raw text. Coverage note: transcript complete 00:00–38:00, auto-generated captions, 2 garbled stretches flagged unclear. Rubric score: 10 out of 10. The map and quotes ship.
Connects securely to your tools. The creator never sees your data.
About this skill
A transcript is raw material, not a result, and the same hour of spoken words can become six different artifacts — each built differently. This skill starts by naming the ask and picking the treatment from a decision table: a layered summary for the gist, a quote pack for lines to reuse or cite, a subtitle file for text on screen, a translation for another language, a fact-check brief for what holds up, or a chapter map for navigating a long recording. Then it cleans the text once — provenance noted, timestamps kept or stripped per treatment, stage directions removed, broken sentences rejoined, garbled stretches flagged as unclear rather than smoothed into invention — and runs the treatment's own protocol. Quote packs cap at 7–10 lines, every quote verified word for word against the raw text with a speaker name and timestamp. Subtitle files break lines for reading at about 42 characters per line and re-time after any edit. Fact-check briefs give one verdict per claim — confirmed, contradicted, or unverifiable — with the source listed, and never hedge with probably true. A five-line delivery rubric finishes the job: an artifact ships at 8 or above out of 10, with every claim traceable to a line of transcript, gaps declared up front, and speaker claims kept separate from analysis. Edge cases cover recordings with no timestamps, transcripts that contradict their own titles, and recordings over two hours; the failure-mode catalog names the title summary, the smoothed garble, the rumor quote, the caption dump, the probably-true verdict, and the single-layer dump.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Translation & Localization.
- It cannot fetch or transcribe a recording; the transcript must already be in hand.
- It cannot verify a quote or a claim from a section that is missing from the transcript; gaps are declared, not filled in.
- A fact-check verdict is only as good as the sources available for the claim; some claims stay unverifiable, and the brief says so.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 3 days ago
Permissions
No special permissions declared or detected