Your data. Your machine. No exceptions.
Retrace is built on a simple principle: your most personal conversations should never leave your computer.
Your data is stored locally. Always.
SQLite database in ~/.retrace/. No accounts, no telemetry.
Conversations, topics, embeddings, reflections are all on your laptop.
When you use Claude or Gemini for reflection, you control exactly what
they can access through labels and permissions.
~/.retrace/retrace.db One file. Your file. Copy it, back it up, delete it. Native macOS, not Electron.
Retrace is built with Tauri v2. It uses a fraction of the memory Electron apps consume, starts faster, and integrates properly with macOS. Standard .dmg install. Nothing unusual in your system.
Open source, forever.
MIT licensed. Read the code, audit the extension, fork it, build on it. Your reflection infrastructure should never be a black box.
The complete data path
ChatGPT, Claude, WhatsApp, Messenger, Instagram, SMS, Obsidian, Calendar, Keep
Extension, drag-and-drop, live sync. All local. No data sent anywhere.
On-device LLM extracts topics, summaries, embeddings. Runs on your hardware.
~/.retrace/retrace.db. One SQLite file on your laptop. That's it.
When data leaves your machine
Most of Retrace works entirely offline: search, network graph, topic extraction, timeline browsing, Life Rivers, Time Capsules display. The only features that send data externally are AI reflection (embedded Claude chat, MCP server) and scheduled automations. This happens under the provider's privacy policy, and you control exactly what they can access.
Used for embedded chat and MCP server. Your existing Claude subscription (Pro/Max). Anthropic does not train on your data. Conversations are not stored beyond the session.
Anthropic Privacy Policy →Used for scheduled reflections (weekly journal, project pulse). Free tier via AI Studio API key. Google does not use API data to train models. No account linking.
Gemini API Terms →Without a Claude subscription or API key, reflection features are unavailable. Everything else works: browse conversations, semantic search, network graph, topic extraction, Life Rivers, Time Capsules, labels, and Obsidian sync. Retrace is still fully useful as a local archive and explorer.
You control what providers see
Retrace has a label-based permission system. Tag conversations with labels (platform, topic, or custom), then set allow/block rules per consumer. The embedded chat and MCP server only see what you explicitly permit.
Claude cannot access conversations tagged "private" or "family"
Claude Desktop can only access your AI conversations, nothing else
Technical details
~/.retrace/retrace.db ~/.retrace/raw/ ~/.retrace/models/