Garage Support Center
Setup instructions, troubleshooting guides, and answers to frequently asked questions for Garage — your local-first personal knowledge retrieval engine. Looking for the app itself? Download Garage.
Explore Support Topics
Getting Started & Setup
Step-by-step setup for both the native macOS menu bar app and the `garage` command line, including source registration and indexing.
View Setup Guide →macOS Permissions & TCC
How to grant Full Disk Access for indexing Apple Messages (chat.db), Mail, and protected system directories without errors.
Configure Permissions →Embedding Models & Providers
Run embedding models on Garage's built-in llama.cpp engine, Ollama or LM Studio. Backfill vectors, distill facts and switch models seamlessly.
Manage Models →Model Context Protocol (MCP)
Connect your personal knowledge index to LLM clients including Claude Desktop, Claude Code, and Cursor via MCP 2.0.
Setup MCP Integration →Troubleshooting & Diagnostics
Resolutions for common issues: PostgreSQL startup errors, port conflicts, DNS rebinding blocks, and placeholder downloads.
Troubleshoot Issues →Frequently Asked Questions
Answers to common questions regarding local storage footprint, supported formats, search algorithms (RRF), and privacy guarantees.
Read FAQ →Privacy & Egress Guarantees
No cloud AI client, one egress choke point with a destination allowlist, and messages that Garage itself never sends off your computer.
Read Privacy Policy →Contact & Bug Reports
How to safely submit diagnostic logs, report bugs, request features, or contact the maintainers on GitHub.
Get Support →Quick Diagnostic Checklist
If you are experiencing unexpected behavior, check these fundamental items first:
- Check Database Status: Ensure PostgreSQL is running. In the macOS App, open the Status page: Health reads “All systems go” or lists each problem with the button that fixes it. In CLI, verify with
garage stats. - Verify Full Disk Access: If indexing
~/Library/Messagesor~/Library/Mail, make sure Full Disk Access is granted to Garage or your Terminal application under System Settings → Privacy & Security. - Verify Embedding Provider: The built-in engine runs inside Garage, so keep the app running for
garage backfilland search; if you use Ollama or LM Studio instead, make sure it is running. - Inspect Application Logs: In the macOS App, open the Logs page to view live streaming logs from PostgreSQL, ingestion, embedding, the MCP HTTP server and the built-in model engine.
For internal pipeline details, schema definitions, and authorship heuristics, check the developer specifications: Architecture Guide, Attribution Engine, and Database Schema Reference.