Log.md

LOG.md is a lightweight work-log primitive designed for AI agents operating within software development workspaces. Agents append entries to a single file as they work, using either JSON or plain markdown format, with no required schema beyond the chosen format. It is part of the AI Dev Workspace Primitives index, a collection of reusable building blocks for organizing agent-driven development activity.

The project's purpose is to provide the simplest possible activity record. It is intended as a starting point: when a project's agent activity grows in volume or complexity, it can graduate to a more structured layout such as a planning/session-logs/ directory. The design deliberately avoids tooling, parsers, or enforced structure, so that the log remains readable by both humans and agents and easy to adopt in any repository.

This primitive may be useful to developers experimenting with autonomous or semi-autonomous coding agents who want a traceable record of agent actions without introducing a dedicated logging system. It also suits teams establishing conventions for agent workflows in early stages, where a plain file offers transparency and low overhead before more formal session tracking is warranted.