plain text, connected

About txt

Why txt exists and how its human and agent surfaces fit together.

txt is a focused writing system built around files people can understand without special software. A note is plain text or Markdown, a folder is a durable organizing boundary, and metadata such as context, tags, links, and versions stays explicit. The product combines a fast web, desktop, and mobile editor with a command-line client, a documented HTTP API, and a Model Context Protocol server. That shared model lets a person write in the interface they prefer while trusted automations work with the same source of truth.

The product is designed for reviewable collaboration. Agents can list before reading, request only the scopes they require, propose or apply bounded changes, and use recoverable trash instead of silently destroying notes. Teams can share folders, publish skills, retain version history, and connect documents without converting their knowledge into an opaque database. Local-first workflows remain useful offline, while optional cloud synchronization makes the selected workspace available across devices and integrations.

txt is operated as an evolving software product with public discovery documents for agents and developers. The OpenAPI specification describes REST operations, llms.txt explains when and how agents should use the service, and the remote MCP endpoint exposes a compact tool surface. Security-sensitive actions require authentication and authorization; public documentation, protocol metadata, and a non-mutating sandbox remain available without an account so integrations can evaluate the platform before receiving access to private data.