Recall

Capture your AI conversations from Claude, ChatGPT, and Gemini as a portable memory layer.

Your AI context shouldn't be trapped in one platform. Recall quietly captures your conversations from Claude, ChatGPT, and Gemini as you chat — no extra steps needed. Messages are stored locally on your device and exported as structured JSONL files you own and control. ▎ Why Recall? Switch models and you lose all your context. Recall gives you a portable memory layer for your AI interactions — capture, search, and retrieve across every platform from one place. ▎ Features - Captures user messages and model responses (your choice) - Works across Claude, ChatGPT, and Gemini - Search your conversation history directly from the side panel - Star and favorite important messages across platforms - Platform-aware search filtering (All / Claude / ChatGPT / Gemini) - Pluggable retrieval backends — ships with MemPalace support for semantic search - Incremental export — only new messages each time - Auto-export every night at 11:59 PM (optional) - Everything stays on your machine — no accounts, no cloud, no third parties ▎ Search Type a query in the Conversations tab and find past exchanges by meaning, not just keywords. Recall connects to MemPalace (https://github.com/MemPalace/mempalace) — a local AI memory system that indexes your conversations with semantic vector search. Search results show the conversation title, a highlighted snippet, and expand to the full thread on click. To enable search, install Recall Bridge: https://github.com/ty-asuralo/recall-bridge ▎ Coming soon - GBrain integration (https://github.com/garrytan/gbrain) — a personal knowledge system with hybrid vector + keyword search, entity linking, and timeline tracking - More retrieval backends — the bridge architecture is pluggable, so new backends can be added without changing the extension - Cross-platform context injection — use your captured memory to prime new conversations on any AI platform ▎ Export format Conversations are saved as JSONL files organized by platform and role, designed to plug into any retrieval or memory layer you build on top.