AI News

note.md: Local-First Markdown Review and Agent Memory on macOS

Originally published: 2026-06-27

Last verified: 2026-07-24

TL;DR: note.md is now presented as a free, open macOS editor for reviewing agent-written Markdown while keeping a person’s highlights, doubts, and decisions in a separate companion file. It uses ordinary folders rather than a cloud database and gives coding agents a simple file-based memory they can read later. That is a narrower and clearer product than the earlier research-workspace description.

What changed in the current product

The June launch coverage described PDF research, citations, semantic search, and paid local-AI workflows. The current official site no longer makes those features or a premium plan the center of the product. It describes a Markdown reader, outliner, daily notes, search, manual links, and marginalia stored beside the source document.

This revision follows the live product rather than retaining claims that the current page does not support. The site says note.md is free and open, requires macOS 13 or later, supports Apple Silicon and Intel Macs, and keeps files on the user’s disk.

The companion-file model

Each document can have a partner file named with the source filename plus a note.md suffix. The source may be regenerated by an agent. The companion file holds the reader’s marks, questions, judgments, and confirmed links. Keeping the two layers separate avoids mixing generated prose with the human record of what mattered.

Daily notes and manually confirmed links can then act as memory for an agent that reads the folder. note.md documents an AGENTS.md file as the place for folder rules and names Claude Code, Codex, OpenClaw, and Hermes as examples of tools that already understand Markdown files.

What local-first does and does not mean

Plain local files reduce lock-in and make backup, versioning, and inspection straightforward. They do not automatically secure the workflow. An agent, sync utility, backup service, or Git remote can still transmit the files. Users must inspect every tool that receives access to the folder.

The current site also does not establish enterprise access controls, managed retention, mobile support, collaborative editing, or a service-level commitment. Teams that need centralized policy and audit features should evaluate those gaps before standardizing on a personal file workflow.

Track later platform changes through AI News, while treating the live repository and release page as the authority for installation details.

Who may find it useful

  • Writers reviewing a large volume of agent-generated Markdown.
  • Researchers who want judgments stored separately from replaceable drafts.
  • Developers who prefer file-based memory that works with several agents.
  • Mac users who want a portable folder instead of a proprietary database.

It is less suitable for Windows or Linux users, teams that need live multi-user editing, or organizations that require centralized identity, policy, and retention controls.

File management matters

The folder is the product’s long-term record, so naming and version control deserve attention. Test what happens when a source file is renamed, moved, duplicated, or replaced by an agent. Confirm that links and companion notes remain understandable without the app. A team using Git should also decide whether generated drafts, personal marginalia, and local indexes belong in the repository or in ignored files.

Evaluation checklist

  1. Install the signed release on a non-sensitive test folder.
  2. Create a source document and confirm how its companion notes are stored.
  3. Check search, links, daily notes, export, backup, and restore behavior.
  4. Give one agent access to the folder and inspect every file it reads or writes.
  5. Verify that human notes survive source regeneration and filename changes.
  6. Document which sync, backup, and agent services can transmit the folder.

Official sources

Related Kingy AI links