Home
Optional sponsorMemory for AI features

Supermemory

Supermemory

Supermemory is a memory engine for AI agents. It stores what each user tells your app, turns it into memories and a profile, and returns the relevant memories when you build the next prompt.

Hackathon perks:To be confirmed

Good to know

  • An AI feature can use what a user said in an earlier session.
  • Each user's memory is kept apart by a namespace, such as their user id.

Get started

  1. 1

    Get a key

    In the console, open API Keys and create one. Keep it on your server, not in the app.

    console.supermemory.ai
  2. 2

    Install the SDK

    npm i supermemory
    export SUPERMEMORY_API_KEY="sm_..."
  3. 3

    Add a memory

    Processing is asynchronous. dreaming: "instant" processes the document now, which is handy while you test.

    import { Supermemory } from "supermemory";
    
    const supermemory = new Supermemory({ apiKey: process.env.SUPERMEMORY_API_KEY });
    const user = "user_4f8a";
    
    await supermemory.add(user, {
      content: conversation,
      id: "chat_2026_11_15",
      metadata: { type: "conversation" },
      dreaming: "instant",
    });
  4. 4

    Search before you prompt

    const results = await supermemory.search(user, {
      query: "what helps them sleep before a morning run",
      limit: 3,
    });
  5. 5

    Set up your coding agent

    Prints an integration prompt you can paste into Capy, and installs the Supermemory skill.

    npx supermemory setup --prompt
    npx skills add https://github.com/supermemoryai/skills --skill supermemory
Ask Capy for memory
Add Supermemory to the chat feature. In a Convex action, add each conversation turn to Supermemory with the user's id as the namespace, and search their memories before building the model prompt. Keep SUPERMEMORY_API_KEY in Convex environment variables, not in the app.
Next