borghei/Claude-Skills / personal-productivity/meeting-insights
meeting-insights
VerifiedAnalyze meeting transcripts to extract decisions, action items, owners, due dates, open questions, and risks. Use after recorded meetings, sales calls, customer interviews, or planning sessions, or to build a decision log.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.3 KB |
| assets/recap_template.md | markdown | 1.5 KB |
| references/insight_extraction_patterns.md | markdown | 3.7 KB |
| scripts/transcript_analyzer.py | script | 6.9 KB |
SKILL.md
View on GitHub---
name: meeting-insights
description: >
Analyze meeting transcripts to extract decisions, action items, owners, due
dates, open questions, and risks. Use after recorded meetings, sales calls,
customer interviews, or planning sessions, or to build a decision log.
license: MIT + Commons Clause
metadata:
version: 1.0.0
author: borghei
category: personal-productivity
domain: meetings
updated: 2026-05-04
python-tools: transcript_analyzer.py
tech-stack: meetings, async-collaboration
tags:
- meetings
- transcripts
- action-items
- decisions
- summaries
---
# Meeting Insights
Turn raw meeting transcripts into a structured set of decisions, action items, owners, due dates, open questions, and risks.
---
## Table of Contents
- [Keywords](#keywords)
- [Quick Start](#quick-start)
- [Core Workflows](#core-workflows)
- [Tools](#tools)
- [Reference Guides](#reference-guides)
- [Templates](#templates)
- [Best Practices](#best-practices)
- [Scope & Limitations](#scope--limitations)
---
## Keywords
meeting, meetings, transcript, notes, minutes, action items, decisions, decision log, follow-up, recap, sales call, customer interview, retrospective, standup, planning, async
---
## Clarify First
Before extracting insights, settle the inputs below. When one is missing or vague and would change the result, ask about it; when it wouldn't, use a sensible default and say which one you used.
- **Transcript with speaker labels:** `Speaker: text` format drives owner attribution on action items.
- **Meeting type:** recap vs customer interview vs decision log changes which extractions matter (decisions/actions vs pains/quotes).
- **Output target:** recap email, append-only decision log, or interview synthesis sets the structure of the deliverable.
Ask about the two or three inputs that matter most for this request, not the whole list. If the user wants a result straight away, produce it and state your assumptions at the top.
---
## Quick Start
### Process a Transcript in 5 Minutes
1. Save your transcript text as `transcript.txt` (one speaker turn per line, format `Speaker: text`)
2. Run:
```bash
python scripts/transcript_analyzer.py transcript.txt
```
3. Review the structured output: decisions, action items, owners, due dates, open questions
4. Drop into `assets/recap_template.md` to send a follow-up
---
## Core Workflows
### Workflow 1: Post-Meeting Recap
**Goal:** Convert a 60-minute conversation into a 90-second readable summary that everyone can act on.
**Steps:**
1. Export the transcript (Otter, Fireflies, Zoom, Google Meet, etc.)
2. Run: `python scripts/transcript_analyzer.py transcript.txt`
3. Verify owners and due dates — the analyzer is heuristic; humans correct
4. Paste structured output into `assets/recap_template.md`
5. Send within 24 hours of the meeting
**Expected Output:** Recap with decisions, action items (owner + due date), open questions, and risks.
**Time Estimate:** 5-10 minutes vs. 30+ for manual note review.
### Workflow 2: Customer Interview Synthesis
**Goal:** Pull the signals out of a discovery call without losing the customer's actual words.
**Steps:**
1. Run analyzer in JSON mode: `python scripts/transcript_analyzer.py transcript.txt --json`
2. Filter for `pains` and `quotes` — these are the discovery signals
3. Use `references/insight_extraction_patterns.md` to triangulate across multiple interviews
4. Tag findings by ICP segment for product / marketing handoff
**Expected Output:** Tagged customer pain list with verbatim quotes per insight.
**Time Estimate:** 15 minutes per interview after the call.
### Workflow 3: Decision Log Maintenance
**Goal:** Build an organizational memory so the same decision is not re-litigated quarter after quarter.
**Steps:**
1. After each meeting, run the analyzer to extract decisions
2. Append to a running decision log keyed by date and topic
3. When a future meeting raises an old topic, search the log first
4. Re-open formally rather than silently overturning
**Expected Output:** Append-only decision log searchable by topic and date.
**Time Estimate:** 2-3 minutes per meeting.
---
## Tools
### transcript_analyzer.py
Reads a transcript text file and extracts:
- **Decisions** — sentences with decision markers ("we decided", "agreed", "going with")
- **Action items** — sentences with action markers ("will", "going to", "by next week"), with heuristic owner + due date
- **Open questions** — sentences ending in "?" or marked with "open question"
- **Risks** — sentences with risk markers ("risk", "concern", "blocker", "if X then Y")
- **Quotes** — distinctive verbatim sentences > 12 words (for customer interview workflows)
```bash
# Human-readable
python scripts/transcript_analyzer.py transcript.txt
# JSON for programmatic use
python scripts/transcript_analyzer.py transcript.txt --json
```
**Transcript format expected:**
```
Alice: We need to decide on the launch date this week.
Bob: I'll send the draft by Friday.
Alice: Are we blocked on legal review?
Bob: Yes, that's the risk — if legal slips, launch slips.
```
---
## Reference Guides
- **`references/insight_extraction_patterns.md`** — Heuristic triggers for decisions, actions, and risks; how to triangulate across interviews
---
## Templates
- **`assets/recap_template.md`** — Post-meeting recap email with placeholder sections
---
## Best Practices
- **Verify before sending.** The analyzer is heuristic; an unverified recap that mis-attributes an action item destroys trust.
- **Owner + date or it does not exist.** An action item without an owner is a hope; without a date, it is a wish.
- **Send within 24 hours.** Memory of who said what fades fast; recap latency directly correlates with action-item completion rate.
- **Quote verbatim.** For customer interviews, the customer's words matter more than your summary of them.
- **Decision log is append-only.** Never silently overturn — re-open with a dated update.
---
## Scope & Limitations
**In Scope:**
- Extraction from a `Speaker: text` transcript of decisions, action items with a heuristic owner and due date, open questions, risks, pains and quotes (`transcript_analyzer.py`).
- A post-meeting recap built from `assets/recap_template.md`.
- Customer interview signals from the pains and quotes output.
- An append-only decision log workflow keyed by date and topic.
**Out of Scope:**
- Recording or transcribing audio. Transcripts come from tools such as Otter, Fireflies, Zoom or Google Meet.
- Tracking follow-through on actions across a recurring meeting (see `project-management/meeting-analyzer/`).
- Synthesis of themes across many customer interviews (see `project-management/discovery/interview-synthesis/`).
- An executive decision register that separates deliberation from approved decisions (see `c-level-advisor/decision-logger/`).
**Important Caveats:**
- Extraction relies on trigger phrases. It misses decisions phrased indirectly and can flag false positives, such as references to past decisions.
- Owners and due dates are guesses. They need checking before a recap is sent.
- Transcripts without speaker labels lose owner attribution.
- The tool does not store the decision log. The user keeps it.
---
## Integration Points
- Pairs with `product-team/agile-product-owner/` for converting interview pains into stories
- Pairs with `project-management/` for action-item tracking
- Feeds into `marketing/` voice-of-customer workflows
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |