What problem does it solve? Manually extracting action items, decisions, and key topics from meeting transcripts is tedious and error-prone. This Skill automates the conversion of raw VTT, SRT, and plain text transcripts into structured, navigable Markdown knowledge packets with full source citations. ## Core Features & Use Cases - Multi-Format Parsing: Deterministic Python-based parsing of VTT files (with LLM fallback for SRT/TXT), including speaker identification, timestamp normalization, and encoding fallback for Windows-1252 and ISO-8859-1 files. - Semantic Extraction: Identifies action items, decisions, questions, topics, and speakers with confidence scores and dual citations (anchors plus timestamps) back to the original transcript. - Domain-Aware Analysis: Nine domain contexts (software engineering, architecture, product management, UX, cloud, security, and more) customize entity extraction for specific professional meeting types. - Mindmap Generation: Produces Mermaid and ASCII mindmaps by default, with quality validation via an integrated critic scoring above a 0.90 threshold. - Use Case: After a two-hour sprint planning meeting recorded in Zoom, run the skill on the exported VTT file to receive an 8-file Markdown packet listing every commitment, blocker, and decision with links to the exact moment each was discussed. ## Quick Start Ask the AI to process your meeting transcript by providing the file path, for example: parse the transcript at /path/to/meeting.vtt and extract all action items and decisions into a structured packet.
