Meeting Minutes Structuring

Problem to solve

Meeting transcripts, shorthand notes, or pasted chat logs are often unstructured: topics, attendees, decisions, owners, and due dates are scattered across long paragraphs. Manual cleanup can miss action items and makes inconsistent phrasing harder to track, such as mixing “as soon as possible” with “by next Friday.”

How the skill works

The skill takes UTF-8 text as input, preprocesses it, and extracts fields according to a structured template:
- Basic information: identifies meeting topic, time, location, and attendees.
- Decisions: locates decision verbs such as decide, approve, or confirm, then outputs the decision content and voting method.
- Action items: extracts owner, task, and due date into an owner | action | due date structure.
- Open issues: groups expressions such as to be discussed, postponed, or pending separately.
Uncertain fields are marked with [needs verification: field] instead of being guessed. It supports default Markdown output and a --format plain plain-text mode.

Boundaries

Input should be Chinese or Chinese-English mixed text, preferably under 500KB per request. It does not process audio or video directly, perform semantic completion, or make subjective judgments. For English-only text, missing fields, or ambiguous wording, verify against the source before sharing the minutes.