meeting-sync

The meeting-sync skill is a specialized productivity tool designed to bridge the gap between your Granola.ai meeting notes and your personal knowledge management system within the PersonalOS architecture. By leveraging the Granola MCP (Model Context Protocol) server, this skill allows users to maintain an organized, searchable archive of meeting transcripts and action items directly in their local Knowledge/Transcripts folder. It is specifically optimized for morning routines and daily planning sessions, ensuring that your recent professional interactions inform your task prioritization and focus items for the day. This agentic skill manages the end-to-end flow of identifying unsynced meetings, presenting them for user verification, and performing the file system operations necessary to persist meeting content securely. It effectively reduces the manual overhead of exporting meeting data, allowing you to stay organized without leaving your terminal or chat-based interface. Key features include: - Automated discovery of unsynced Granola meetings via MCP integration. - Flexible sync options, allowing for bulk operations or selective imports of specific meetings. - Seamless integration with existing PersonalOS flows like morning planning and task list updates. - Automated sanitization of meeting filenames to ensure local file system compatibility and organization. - Persistent sync state tracking via the .granola-sync.json configuration file to prevent duplicates. Practical usage notes and constraints: - This tool is designed primarily for use with the Granola.ai macOS application and requires the Granola MCP server to be correctly configured. - Ensure your environment path is set correctly in .mcp.json to allow the agent to reach the Knowledge directory. - The skill prioritizes content-rich meetings and may ignore placeholder entries labeled as (no notes). - Users are encouraged to perform syncs during morning stand-ups or when reviewing backlog progress to maintain high-quality context for their AI assistant. - Files saved are plain-text/markdown format, perfect for future processing by LLMs for pattern analysis or summarization.