Whether you're new to md-log or rolling it out to a team, everything you need is here — from the overview to connecting your agent, reviewing & annotating, and self-hosting.
md-log is a human-in-the-loop review and archive layer where an AI coding agent writes its work as Markdown (.md) reports over MCP, and a person reviews them on web, mobile, or tablet, annotating with a pen, while every save becomes an immutable version.
You can start md-log in about 5 minutes with three steps: create a free account, mint an MCP key in Settings → Tokens, and connect your AI coding agent with npx -y md-log-mcp. Then review the .md reports your agent saves on web, mobile, or tablet.
md-log's MCP server connects Claude Code, Claude Desktop, OpenAI Codex, Cursor, and other MCP clients with one command (npx -y md-log-mcp) and two environment variables — the service URL (https://app.md-log.com/api/v1) and your MCP key (mdlog_pat_...) — so your AI agent can save its work reports to md-log as Markdown.
md-log is the review and archive surface for your AI coding agent's work: the agent writes Markdown reports by path into a folder tree, and you read, approve or reject, and hand-annotate them on web, phone, or tablet — with every save kept as an immutable version.
Team collaboration in md-log lets you create a team and share a document or a whole folder with teammates so you can review reports together. It is free and fully hosted, capped at 5 members per team; you invite teammates by email (they need an existing md-log account), assign roles from Viewer to Owner, and everyone sees each author's pen annotations pinned to the version. Need more than 5 members or data behind your firewall? Move to On-Premise.
