Loreqo Keeps AI Coding Agents Aligned

Field notes

The workflow begins with an Obsidian ZIP or Markdown folder rather than a new proprietary documentation format. Loreqo says it extracts project knowledge through the user's own AI provider account, then exposes relevant decisions through MCP-connected agents. Its example project contains 318 imported documents and 47 tracked decisions or constraints, including a row-level-security violation. A documentation-folder watcher surfaces changed sections and proposed constraints for review while source code remains outside the synchronized material.