Annot is a personal knowledge base that doubles as the memory layer under Claude, ChatGPT, Cursor and Gemini. Anything you save - a webpage, a PDF, a voice memo, a note - gets extracted, transcribed or OCR'd as needed and embedded into one searchable store. Over MCP your AI tool can search it, cite it, and write back to it. Most memory servers store facts the agent decides to keep. Annot starts from the other end: the library you already built. Your agent doesn't just remember what it was told - it can read what you read. **29 tools** - **Search and recall (5)** - semantic and keyword search over saved sources, image search, find-related, find-similar-images - **Read (7)** - fetch a source, a topic, your focus; list recents, tags, topics - **Working context (4)** - load your current focus into a session, build a context pack for a task, load knowledge or a skill - **Write-back (6)** - agents save links, documents, their own notes and your thoughts; update sources and focus - **Organize (3)** - add, remove and move items between topics - **Export (3)** - generate, fetch and delete context artifacts - **Focus (1)** - answer a focus prompt Requires an Annot account: OAuth runs against your Annot login. Early access at useannot.com.
First seen 2 Oct 2026. One server, whatever directories list it: each directory listing keeps its own page and history.
Tools
No tool list captured yet.
| Directory | Listing | Tier | First seen |
|---|---|---|---|
| Smithery | Annot | - | 2 Oct 2026 |