---
title: Memanto
description: Recall relevant long-term context and let agents deliberately save durable user memories.
url: "https://eve.dev/integrations/memanto"
docs_index: /llms.txt
type: memory
keywords:
  - memory
  - long-term memory
  - semantic search
  - automatic recall
  - user profile
  - Memanto
  - Moorcheh
---

> For an index of all documentation, see [/llms.txt](/llms.txt).

Memory provider integration for eve. Recall relevant long-term context and let agents deliberately save durable user memories.

## Install

Install Memanto memory for eve:

```bash
eve add memory/memanto
```

This installs `@moorcheh-ai/memanto` and writes a memory slot. The provider requires Node.js 24 or later and eve 0.60.0 or later.

## Quick start

Set a Moorcheh API key in the agent environment. Deployments also need the URL of a Memanto server that you run:

```bash title=".env.local"
MOORCHEH_API_KEY=...
# Deployments only
MEMANTO_BASE_URL=https://...
```

The registry creates this memory slot:

```ts title="agent/memory/memanto.ts"
import { memantoMemory } from "@moorcheh-ai/memanto/eve";
import { defineMemory } from "eve/memory";
import { byPrincipal } from "eve/memory/scope";

export default defineMemory({
  description: "Recall and manage durable context for the current user.",
  provider: memantoMemory({
    apiKey: process.env.MOORCHEH_API_KEY,
    baseUrl: process.env.MEMANTO_BASE_URL,
    capture: false,
  }),
  scope: byPrincipal,
});
```

The filename creates the `memanto` slot. Before every turn, the provider recalls relevant context. It also gives the model `memanto__remember` and `memanto__recall` tools for deliberate saves and searches.

## Configure

`byPrincipal` keeps memory disabled for anonymous and runtime principals, and shares the local-development scope while you run `eve dev`. For a multi-tenant agent, replace it with a scope resolver that derives both tenant and caller identity from verified session context. See [Multi-tenant memory](/docs/patterns/multi-tenant-memory).

All eve scopes share one Memanto agent. The provider tags each memory with a digest of eve's opaque scope key and filters recalls by that tag. Its tools are bound to the current scope, so the model cannot read or write another scope's memories.

Automatic capture is unsupported in this reviewed configuration. The published adapter does not make completed-turn capture idempotent across eve operation replays. The registry explicitly sets `capture: false`; leave it disabled until the provider supports replay-safe idempotency. The model can still use `memanto__remember` for deliberate, scope-bound memory writes. Those are model-invoked writes, not user-approved writes. Recalled content is user-provided data, not instructions. Do not store secrets or credentials in memory.

The reviewed configuration expects a running Memanto server at `MEMANTO_BASE_URL`. If you omit it during `eve dev`, the adapter starts a local server with the unpinned default `memanto` package through `uvx`. That local-server path is outside the reviewed compatibility boundary. Keep `MOORCHEH_API_KEY` in the environment rather than prompts or source control. See the [Memanto eve guide](https://docs.memanto.ai/integrations/eve) for server setup and all provider options.

[Read the full memory provider documentation](https://docs.memanto.ai/integrations/eve)

---

For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)

For an index of all available documentation, see [/llms.txt](/llms.txt)

For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)