Agent skill · software engineering · jeremylongshore
granola-local-dev-loop
Access Granola meeting data programmatically for developer workflows.
Why this skill is useful
Adds executable scripts for accessing and processing Granola meeting data that aren't in public documentation.
What it needs
Requires granola, node, python3 installed locally. Requires granola account access. About 4k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Granola Local Dev Loop Overview Access Granola meeting data programmatically using three methods: the local cache file (zero-auth, offline), the MCP server (AI agent integration), or the Enterprise API (workspace-wide access). Build developer workflows that turn meeting outcomes into code tasks, documentation, and project artifacts. Prerequisites Granola installed with meetings captured Node.js 18+ or Python 3.10+ for scripts For MCP: Claude Code, Cursor, or another MCP-compatible client For Enterprise API: Business/Enterprise plan + API key Instructions Step 1 — Read the Local Cache (Zero Auth) Granola stores meeting data in a local JSON cache file: The cache has a double-JSON structure (JSON string inside JSON): Step 2 — Set Up Granola MCP Server Granola's official MCP integration connects meeting context to AI tools: With MCP connected, Claude Code and Cursor can: Search across all your meetings by topic or person Pull context from specific meetings into coding sessions Create tickets based on discussed bugs or features Scaffold code based on architectural decisions from meetings Community MCP servers with additional features: pedramamini/GranolaMCP — CLI + programmatic + MCP access, reads local cache mishkinf/granola-mcp — semantic search with LanceDB vector embeddings proofgeist/granola-mcp-server — lightweight local cache reader Step 3 — Extract Action Items to Dev Tools Step 4 — Sync Meeting Outcomes to Project Docs Step 5 — Git Integration Pattern Reference Granola meetings in commits and PRs: Output Local cache accessible for offline meeting data reads MCP server connected for AI-assisted meeting context Action item extraction pipeline ready Meeting-to-dev-tools sync established Error Handling Error Cause Fix ------- ------- ----- Cache file not found Granola not installed or never launched Install Granola and capture at least one meeting JSON parse error Double-JSON structure not handled Parse the outer string first, then parse the inner object MCP server …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/granola-local-dev-loop