Smart Meeting Minutes Generator

The Problem

Raw meeting materials are often transcripts, handwritten notes, or semi-structured records where context is easy to lose: participants are referenced by pronouns, deadlines are vague, implicit commitments are not listed as actions, repeated topics are scattered, and unresolved disagreements only appear inside the conversation. Manual cleanup can miss owners, invent dates, or turn unconfirmed issues into decisions. @user_f37e97ba/smart-meeting-minutes addresses this by turning meeting text into a traceable minutes structure.

How It Works

The skill processes input through a constrained workflow rather than producing a loose summary:
- Preprocessing: It identifies the meeting type and extracts the meeting name, date/time, and participant list, marking missing details instead of fabricating them.
- Topic breakdown: It groups the content into 3-7 topics, each with a short title and a 2-3 sentence summary of the core discussion.
- Decisions and actions: It separates consensus decisions from follow-up work. Action items should include owner, deadline, priority, and status.
- Disagreement tracking: It records unresolved issues in a table with topic, competing views, and current status, making open questions easier to follow.
- Quality checks: The referenced validate.sh and report-template.md resources are used to verify completeness and template consistency.

Boundaries

It is suited for generating structured minutes from existing meeting text, not for replacing recording, transcription, or live meeting participation. When names, owners, or deadlines cannot be confirmed, the skill requires explicit markers such as to be confirmed or [speaker unknown] and avoids guessing dates. A single processing limit is 10,000 characters, so longer materials should be split. Output language should match input language, so Chinese input normally produces Chinese minutes.