MCP Servers · silverstein

minutes

Minutes is an open-source, privacy-first conversation capture tool that records meetings and voice memos, transcribes them locally using Whisper, and makes them searchable through a relationship graph. It stores everything as plain markdown files in ~/meetings/ and integrates with AI agents via MCP protocol, allowing queries like 'what did I promise Sarah?' across your entire conversation history.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/silverstein/minutes

1.3k

GitHub stars

140

Forks

Rust

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	silverstein/minutes
Owner	silverstein
Primary language	Rust
License	MIT — OSI-approved
Stars	1.3k
Forks	140
Open issues	28
Latest release	v0.19.0 (2026-06-25)
Last updated	2026-07-07
Source	https://github.com/silverstein/minutes

What minutes is

Written in Rust, Minutes combines local audio capture, Whisper-based transcription, native speaker diarization (pyannote-rs), and optional cloud LLM summarization. It exposes a structured markdown format with YAML frontmatter, SQLite indexing for relationships and commitments, MCP tools for agent integration, and CLI/desktop surfaces. All audio processing stays local unless cloud summarization is explicitly enabled.

Quickstart

Get the minutes source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/silverstein/minutes.git2cd minutes3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

AI agent memory layer

Use Minutes as a queryable conversation context store for Claude Code, Codex, Gemini CLI, or other MCP-compatible agents. Agents can search meetings, extract decisions, and recall commitments without external APIs or vendor lock-in.

Local-first meeting notes and recall

Organizations or individuals prioritizing data privacy can record, transcribe, and search meetings entirely on-device. All artifacts are plain markdown in a local directory, readable by grep, git, and any AI tool with file access.

Voice memo and call recording workflow

Capture phone calls, Zoom/Teams/Meet sessions, or ambient voice memos, with desktop app native audio capture (macOS 15+) and auto-processing. Integrates with Obsidian/Logseq vaults for knowledge management pipelines.

Implementation considerations

Audio device routing for call recording (Zoom, Teams) requires manual setup (BlackHole + Multi-Output Device on macOS, or native ScreenCaptureKit on macOS 15+); document this for end users.
Whisper model download (466 MB 'small' model recommended) is a one-time setup step; plan for bandwidth and disk space in deployment.
Optional cloud LLM summarization (Claude, Ollama, Mistral, OpenAI) requires API keys; decide whether to support all or restrict to on-device summarization.
Frontmatter schema and relationship graph structure are custom; invest in documentation or templates if teams will query or ingest Minutes artifacts programmatically.
Consent disclosure mode requires user interaction and decision-making; if mandated by policy, ensure UI/UX is clear and audit trails record user choices.

When to avoid it — and what to weigh

Need real-time multi-party synchronization — Minutes is single-machine-centric. No built-in cloud sync, shared team workspaces, or collaborative real-time transcription. Each user owns their own ~/meetings/ directory.
Require compliance-grade audit trails or encryption — No mention of encryption at rest, key management, audit logging, or formal security certification. Privacy relies on local storage; if compliance demands encrypted backups or tamper evidence, requires custom integration.
Expect end-to-end polished UI or extensive integrations — Desktop app and web UI exist but project focuses on CLI/MCP tooling. Integrations with CRM, calendar, or enterprise systems are not documented. Frontmatter schema is custom and requires custom parsing.
Windows-first or non-macOS deployment — Primary surface is macOS (brew, native capture via ScreenCaptureKit). Windows CLI install requires Rust + LLVM. Desktop app availability on Windows/Linux not stated. MCP server (npx) is cross-platform but native capture is macOS-only.

License & commercial use

MIT License. Permissive OSI-approved license allowing commercial use, modification, and redistribution with attribution and no warranty. No copyleft; proprietary or closed-source derivatives are permitted.

MIT permits commercial use without restriction. However, 'commercial use' of the software itself (selling Minutes or a Minutes fork) differs from *using* Minutes to build a commercial application (e.g., agent-as-a-service). The latter is clearly allowed. The former should be reviewed with legal counsel if you plan to resell or rebrand Minutes directly. No patent indemnity is stated in MIT.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Good
Assessment confidence	High

Security considerations

All audio transcription, diarization, and optional summarization can run locally—no data leaves the machine unless cloud LLM summarization is explicitly enabled. Audio files and transcripts are stored as plain markdown in ~/meetings/, subject to standard OS file permissions. No mention of encryption at rest, secure deletion, or key management. If audio files are deleted, standard OS recovery tools may recover them. For sensitive meetings, there is a 'capture: none' mode (no audio file) with typed notes. Consent disclosure is a UX aid, not a cryptographic proof. No formal security audit, threat model, or incident response process is documented.

Alternatives to consider

Otter.ai

Cloud-hosted transcription and meeting notes with AI summaries and search. Handles team sharing and multi-device sync. Trade-off: vendor lock-in, ongoing subscription, audio uploaded to Otter's servers.

Fireflies.io

AI meeting assistant with real-time transcription, speaker identification, and integrations (Slack, Zapier, CRM). Built for team collaboration. Trade-off: cloud-dependent, less control over data, higher cost at scale.

Obsidian Sync + local Whisper (homebrew)

DIY approach using Obsidian as the front-end, custom scripts to run Whisper locally, and manual relationship graph curation. Cheapest for privacy-first teams. Trade-off: requires custom integration, no out-of-box MCP support, higher maintenance burden.

Software development agency

Build on minutes with DEV.co software developers

Start with the macOS desktop app (brew install --cask silverstein/tap/minutes) or explore MCP integration for your AI workflows. All audio stays on your device.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

elevenlabs-mcp basic-memory MiniMax-MCP

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI application development API development Custom software Talk to DEV.co

minutes FAQ

Does Minutes upload my audio to the cloud?

No, by default. Transcription (Whisper), diarization, and storage all happen locally. If you enable cloud LLM summarization (Claude, Ollama, OpenAI), only the transcript text is sent; the audio file stays on your machine. You control this per-recording.

Can I use Minutes in a team or shared workspace?

Not directly. Each user has their own ~/meetings/ directory. Minutes is designed for individual or agent-centric workflows, not real-time team collaboration. You can sync files to a shared repo or vault, but there is no built-in conflict resolution or multi-user locking.

What LLMs can I use for summarization?

Claude, OpenAI, Mistral, Ollama, and others. You provide API keys or run Ollama locally. The README and docs do not list an exhaustive integration matrix; you may need to test or contribute if your LLM is not listed.

Is there a mobile app?

Phone voice memos can be synced to your Mac and processed via Minutes CLI. A dedicated mobile recording app is not documented. Desktop app is macOS-only; CLI and MCP server are cross-platform.

Custom software development services

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If minutes is part of your mcp servers roadmap, our team can implement, customize, migrate, and maintain it.

Ready to own your conversations?

Start with the macOS desktop app (brew install --cask silverstein/tap/minutes) or explore MCP integration for your AI workflows. All audio stays on your device.

Talk to DEV.co Schedule a Consultation