Local-first conversation memory app with AI agent integration via MCP.
Transcribe meetings and voice memos on-device into Markdown files your AI agents can query.
Minutes is an open-source conversation memory app that records and transcribes meetings, calls, and voice memos directly on your device. It stores records as local Markdown files with YAML frontmatter, letting MCP clients like Claude Code, Cursor, and Codex search your history. It is built for agent-first professionals who want a local capture layer for their second brain.
The application handles on-device audio capture, local transcription, and speaker processing, saving output into ~/meetings/ as structured Markdown files. It features a desktop app with dictation preferences, a CLI with 58 commands, an MCP server with 34 tools, and a Claude Code plugin with 23 skills. Text only leaves your machine when you explicitly send authorized meeting context to your chosen cloud AI assistant or summarizer.
It suits agent-first users who want to query their meeting history through Claude Code, Cursor, or the CLI and fold context into a local vault or wiki. It does not suit users looking for a fully managed, cloud-hosted workspace or those who want an out-of-the-box cloud transcription service without local setup.
Transcribes meetings, calls, and voice memos locally on your hardware.
Exposes 34 local tools to query your meeting history via Claude Code, Codex, or Cursor.
Saves all records as plain Markdown files with structured YAML frontmatter you own.
Provides menu bar capture, custom spellings, snippets, and local text recovery.
Minutes is free and open source under the MIT license, with no paid tiers for anything in the repository. The only cost is running it yourself and any external API usage if you choose a cloud AI assistant for summarization.
876 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
All transcriptions are saved as plain Markdown files with YAML frontmatter in your local ~/meetings/ directory.
Yes, it supports any MCP client including Cursor, Codex, OpenCode, Pi, Gemini CLI, and Claude Desktop.
No, transcription and speaker processing run entirely on-device, and audio stays on your machine.
See Zekai first in Google