Recall
Long-term memory for AI agents, visible to humans
Start new thread

Recall - Long-term memory for AI agents, visible to humans

by•
Recall is a local-first memory layer for AI agents: Markdown for humans, SQLite for search, MCP for agents. Agents write structured durable facts through CLI/MCP/API. Humans browse, search, inspect, and audit those memories through a read/view-first UI. The source of truth stays as Markdown on your machine; SQLite is a rebuildable local index. Agents write. Humans inspect. Recall indexes. MCP retrieves.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt 👋 I built Recall because I was tired of losing context when switching between laptops, projects, or different AI agents. Every tool seemed to store memory differently, and most of the time I had no clear way to see what was saved, edit it, or understand what the AI would remember later. Recall is a shared memory layer for both humans and LLMs. It includes an MCP server, so compatible AI tools and agents can store, search, and update long-lived memories through a standard interface. humans can review and search memories in one UI AI agents can use MCP or CLI to access memory across sessions context stays portable instead of being locked inside one chat, laptop, or tool The goal is simple: make AI memory visible and reusable. Would love your feedback, especially from people building with multiple agents, MCP tools, or multi-device workflows.