meeting-lens 0.1.0

CLI and library for extracting structured meeting summaries from transcripts.

meeting-lens

meeting-lens is a Python CLI and library that turns raw meeting transcripts into structured summaries: decisions, action items, open questions, and sentiment.

Why this exists

Meeting notes are painful for engineering managers and team leads: long transcripts are hard to skim, key decisions get buried, and follow-ups are easy to miss. meeting-lens converts transcript exports from common tools into concise, structured outputs that are easy to share or automate.

Quick start

1) Install

pip install "meeting-lens[openai]"

Or install all backends:

pip install "meeting-lens[all]"

2) Set an API key

export OPENAI_API_KEY="your-key"
# or
export ANTHROPIC_API_KEY="your-key"

3) Run your first summary

meeting-lens transcript.vtt

CLI usage

# Basic usage
meeting-lens transcript.vtt

# Specify backend and output format
meeting-lens transcript.txt --backend anthropic --format markdown

# Write to file
meeting-lens transcript.srt --backend openai --format json --out summary.json

# Override model
meeting-lens transcript.vtt --model gpt-4o-mini

# Pipe support
cat transcript.txt | meeting-lens - --backend anthropic

Options:

• --backend: openai | anthropic (default: openai)
• --format: markdown | json (default: markdown)
• --out: output file path (default: stdout)
• --model: override backend default model
• --no-color: disable ANSI color in markdown output

Supported input formats

Format	Extension	Notes
Plain text	.txt	Supports lines like Speaker: message
WebVTT	.vtt	Works with Zoom/Teams-style VTT exports
SubRip	.srt	Parses subtitle blocks and speaker prefixes

Backend comparison

Backend	Package extra	API key env var	Default model
OpenAI	meeting-lens[openai]	OPENAI_API_KEY	gpt-4o
Anthropic	meeting-lens[anthropic]	ANTHROPIC_API_KEY	claude-sonnet-4-20250514

Markdown output example

# Meeting Summary

**Date:** 2024-03-15  
**Participants:** Alice, Bob, Carol  
**Sentiment:** 🟡 Mixed — _Some tension around timeline but positive on outcome._

## TL;DR
Two-sentence summary of the meeting.

## ✅ Decisions
- **Alice:** We will migrate to the new auth system by Q2.
- **Team:** Daily standups move to 09:30.

## 📋 Action Items
| Owner | Task | Due |
|-------|------|-----|
| Bob | Write migration runbook | Friday |
| Carol | Schedule design review | — |

## ❓ Open Questions
- What is the rollback plan if migration fails? _(raised by Bob)_
- Who owns the staging environment credentials?

## 📊 Sentiment
**Mixed** — Brief explanation of sentiment rating.

Library usage

import asyncio
from meeting_lens.backends.openai_backend import OpenAIBackend
from meeting_lens.formatters.markdown import MarkdownFormatter
from meeting_lens.lens import MeetingLens

async def run() -> None:
    transcript = "Alice: We should ship Friday.\nBob: I will own rollout docs."
    lens = MeetingLens(backend=OpenAIBackend())
    summary = await lens.run(transcript, source_name="sync.txt")
    print(MarkdownFormatter().format(summary))

asyncio.run(run())

Contributing

1. Fork and clone the repo.
2. Create a virtual environment and install dev dependencies.
3. Run tests with pytest.
4. Open a PR with clear commit messages and test notes.

Publishing (maintainers)

meeting-lens is published to PyPI using standard Python packaging flow:

python -m pip install --upgrade build twine
python -m build
twine check dist/*
twine upload dist/*

Notes:

• Set TWINE_USERNAME=__token__
• Set TWINE_PASSWORD to your PyPI API token
• Always bump version in pyproject.toml before upload

Related

• context-trimmer
• mobile-context-trimmer