Convert meeting transcripts into detailed retellings and day-level summaries.
npx skills add https://github.com/alatyshau/duet --skill digestor-alatyshau
Please help me install this Agent Skill. Skill: digestor Source: https://github.com/alatyshau/duet/tree/main/packages/skills/tools/digestor Command: npx skills add https://github.com/alatyshau/duet --skill digestor-alatyshau
retelling.md with attributed, third-person prose organized by thematic blocks. - Day-Level Summary: Aggregates all retellings from one day into a single summary.md at the meeting folder level (stage defined, format pending calibration). - Parallel & Long-Transcript Handling: Processes multiple transcripts in parallel agents and splits oversized transcripts into parts, merging results without losing content at the seams. - Use Case: After a day of recorded meetings, ask the agent to retell each transcript; you get one retelling.md next to every transcript.md, preserving decisions, disagreements, and explanations with speaker attribution. ## Quick Start Ask the agent to run the digestor on today's meeting transcripts to generate a detailed retelling for each recording.High-intent search queries and answers about installing and using this skill.
Run the digestor pipeline on the transcript file. It reads the full transcript, classifies content into meaningful material versus noise, and writes a retelling.md with thematic sections and third-person attributed prose next to the original file.
Point the skill at all transcripts from the day. Each transcript is retold in a separate parallel agent producing its own retelling.md, then a day-level summary.md can be generated across all retellings in the meeting folder.
Fillers, technical interruptions, lateness coordination, household logistics, and content-free repetitions are excluded. Decisions, tasks, arguments, explanations, and contextual remarks are kept; when in doubt, content is included rather than dropped.
Yes. Long transcripts are read in parts, each part is retold separately, and the partial retellings are merged into one file. The process watches chunk boundaries to avoid losing content or creating duplicates.
No. The user decides when to run each stage. The agent does not move from retelling to summary on its own, so each compression step is explicitly triggered.
jaimedhenriquesTranscribe audio and video recordings into text with optional speaker diarization.
NousResearchExtract decisions, owners, and action items from meeting transcripts and notes.
gnektGenerate structured meeting briefs from emails, notes, and vault references.
onebrain-aiFetch web pages and save structured markdown summary notes with frontmatter.