# Second Brain Second Brain gives connected AI tools a shared memory for projects, decisions, preferences, and updates. It runs in your own Cloudflare account and lets you save, recall, inspect, and organize context across supported clients and capture sources. - Findings24: - ウェブサイト: - 料金: Freemium - カテゴリー: AIエージェント基盤, AIナレッジベース ## Second Brainとは？ Second Brain is a self-hosted shared-memory layer for AI tools. It lets connected clients save useful context—such as project decisions, preferences, and open threads—and retrieve that context in a later conversation. Instead of maintaining separate memory in each tool, clients connect to the same brain through a shared service. The system supports semantic recall, exact search, source information, and time history. Memories can be updated, linked, organized with tags, and grouped into projects. Retrieval can return related entries and synthesized answers with supporting memories, while source and staleness information helps users check whether recalled information is still current. Second Brain can be deployed to a user’s own Cloudflare account through its desktop app, which is available for Mac and Windows. The app helps provision the Worker and connect clients; Claude Code and Cursor have guided setup, while other clients can use a connection URL and supplied instructions. The implementation is open source, and the architecture uses Cloudflare Worker, D1, Vectorize, Workers AI, and KV components. The product also supports capture beyond AI chat. Users can save from the dashboard, terminal, share sheet, bookmarklet, Obsidian, and other configured sources, while calendar and inbox synchronization are described as read-only. Projects provide named homes for memories, and team usage includes private personal layers alongside an intentionally shared layer. ## Second Brainでできること ### Shared memory across AI clients A single MCP-connected service lets supported tools read and write the same brain, so context saved in one client can be available in another. ### Semantic and exact recall Users can ask for the meaning of a saved item or search for exact names, ticket numbers, version strings, and phrases. Full-text search and semantic retrieval serve different lookup needs. ### Sources, history, and staleness context Memories can retain their source and time history. Recalled answers list supporting memories, and staleness hints help users review information that may no longer be current. ### Memory maintenance and relationships Second Brain checks for duplicates, supports updates and replacements, and can connect related entries. Newer information can take precedence in an answer while older records remain in the history. ### Projects and flexible organization Named projects, tags, and tag aliases help separate contexts such as a website, client, or household project. Existing tags can be associated with a project without rewriting the underlying notes. ### Multiple capture and team layers Context can be captured through several interfaces and configured sources. Team Edition provides a private layer for each person and a shared layer populated only when someone deliberately shares a memory. ## 利用シーン ### Moving work between AI tools Explore an idea in one client, then continue implementation in another without manually repeating the project background, decisions, or constraints. ### Tracking project decisions Save a launch decision, design constraint, or open thread and later ask where a project stands. Project grouping narrows recall to the relevant body of work. ### Maintaining personal working preferences Store writing preferences, recurring instructions, and lessons learned so connected tools can use them in future conversations. ### Capturing context where it occurs Save an idea from a terminal, share sheet, bookmarklet, dashboard, or Obsidian, while configured calendar and inbox sources can provide read-only updates. ### Separating private and shared team knowledge Keep personal memories private while intentionally sharing selected entries with a team’s common layer. Administrators manage members and roles but cannot read private memories. ## よくある質問 ### Does connecting a client import its entire chat history? No. Connecting a tool gives it access to the brain but does not import that tool’s complete chat history. Automatic capture depends on the client following its instructions or on a configured integration. ### How does Second Brain return an answer? A connected client sends a recall request. Second Brain retrieves relevant memories by meaning or exact text, may follow relationships to linked entries, and can return synthesized context with the supporting memories listed. Retrieval is a starting point rather than a guarantee that every answer is complete or current. ### Where is the data deployed? The desktop app deploys Second Brain into the user’s Cloudflare account. Its documented architecture uses a Cloudflare Worker for MCP and REST requests, D1 for entries and relationships, Vectorize for semantic retrieval, Workers AI for model inference, and KV for authentication-related state. ### Can teams use Second Brain without exposing private memories? Yes. Each team member has a private layer that other members cannot read, search, or reach through a client, plus a shared layer for memories deliberately shared with the team. The site states that Team Edition runs on the same Worker at no extra charge and that there is no paid team tier. ## クイック情報 - **Product type**: Shared AI memory and context-management tool - **Deployment**: User’s own Cloudflare account - **Desktop platforms**: Mac and Windows - **Connection model**: MCP and REST; clients use one shared service - **Source and license positioning**: Open-source software - **Notable sources**: Dashboard, terminal, share sheet, bookmarklet, Obsidian, calendar, and inbox integrations