meeting-scribe

Generates decision records, filtered follow-up drafts, and unresolved-item lists from recorded Littlebird meetings.

84|37|Updated May 23, 2026
One-click install
npx skills add https://github.com/legioncodeinc/vibe-coding-tools --skill meeting-scribe-legioncodeinc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: meeting-scribe
Source: https://github.com/legioncodeinc/vibe-coding-tools/tree/main/src/quarantine/skills/keep/meeting-scribe
Command: npx skills add https://github.com/legioncodeinc/vibe-coding-tools --skill meeting-scribe-legioncodeinc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Built-in meeting summaries are written for you, not for the people you met with, and forwarding them leaks private notes, internal observations, and pricing. Decisions get disputed later because nobody kept the exact wording, and questions that got talked over quietly die. This Skill turns one recorded Littlebird meeting into a sendable follow-up, an evidence-backed decisions record, and a list of what never landed. ## Core Features & Use Cases - Recipient-aware follow-up drafts: Builds outbound recaps filtered for confidentiality before drafting, splits drafts by audience when multiple parties were in the room, and holds everything for your approval. It never sends. - Decision records with receipts: Anchors each decision to the exact transcript quote and timestamp, with status (standing, superseded, contested), so a disputed "we agreed X" is settled by evidence. - Unresolved detection: Runs a three-tier sweep covering acknowledged open questions, explicitly parked items, and talked-over questions found by scanning the transcript. - Evening sweep routine: Creates a daily 18:00 Littlebird routine that flags meetings needing follow-ups and escalates ones flagged three times. - Use Case: After a client call ends, ask for the recap and get a filtered draft addressed to the client, a decisions record with quotes, and a list of questions nobody answered. ## Quick Start Ask your assistant to write the follow-up for the call that just ended, and approve the drafted recap before anything is sent.

Frequently Asked Questions about meeting-scribe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a follow-up email after a recorded meeting?▼

Ask for the follow-up right after the call ends. The skill pulls the meeting summary and attendee list from the Littlebird MCP, filters out private and internal content, drafts a recipient-aware recap, and holds it for your approval before anything is sent.

How do I prove what was decided in a meeting?▼

The skill builds a decisions record where each entry carries the verbatim decision, the decider tag from the summary, the exact transcript quote, and its timestamp. A disputed decision is settled by the recorded evidence rather than by memory.

Does meeting-scribe send the recap email automatically?▼

No. The follow-up is always a draft held for approval, even when an email connector is present in the session. Without a connector, you get a copy-paste block with the recipient list and subject line.

What are the requirements for using Littlebird meeting skills?▼

You need the Littlebird MCP connected on a Power or Pro plan, plus a recorded meeting. A linked calendar event is required for recipient-aware drafting; without an attendee list, the decisions record and unresolved list still run but drafting cannot.

What happens if a meeting has no linked calendar event?▼

The skill reports that no attendee list exists and skips recipient-aware drafting rather than inferring attendees from the transcript. It still produces the decisions record, unresolved list, and commitment handoff, and offers to draft against a user-supplied recipient list.

Can it catch questions that were asked but never answered?▼

Yes, through a three-tier unresolved scan covering acknowledged open questions, explicitly parked items, and talked-over questions found by reading forward from each interrogative in the transcript. Tier 3 results carry confidence ratings and stay internal by default.