Meeting Notes

Meeting Notes Skill

Process Microsoft Teams meeting transcripts (VTT format) to extract meeting minutes, action items, and decisions.

Task Flow

1. Transcript Parsing: Read VTT (WebVTT) format files. See [references/formats.md](references/formats.md) for parsing logic.
2. Analysis: Identify decisions, actions, and key discussions using NLP and pattern matching.
3. Generation: Produce structured markdown meeting minutes. See [references/examples.md](references/examples.md) for standard output format.
4. Integration: Offer to update GitHub issues, Jira tickets, or Slack messages.

Output Structure

The primary output is markdown minutes containing:

• Meeting Header (title, date, attendees, duration)
• Executive Summary (1-2 sentence overview)
• Decisions (bulleted list with rationale and owners)
• Action Items (table format with owner, description, and due date)
• Key Discussion Points (organized by topic)
• Blockers & Follow-ups (open questions, escalations)

Reference Materials

• Formats & Logic: [references/formats.md](references/formats.md) - VTT format details, parsing logic, and system integration commands.
• Examples: [references/examples.md](references/examples.md) - Comprehensive before-and-after demonstration of a meeting processing workflow.

Features

• Automatic Detection: Identify speakers, decisions ("we agreed"), and actions ("@owner will").
• Confidence Scoring: Heuristic-based scoring for decision certainty and action clarity.
• System Integration: Automated gh issue comment, jira issue create, and slack chat postMessage.

Related Skills & Tools

• github-cli - For posting to GitHub issues
• jira - For creating Jira tickets
• slack-api - For posting to Slack channels
• markdown - For formatting output

Troubleshooting

• Poor Transcript Quality: Manually provide speaker list in the first line of the VTT.
• False-Positives: Use the summary prompt to manually filter incorrect actions before integration.
• Large Meetings: Be concise in discussion summaries to avoid context bloat.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: shalomb
• Source: shalomb/agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.