Skip to main content

In-Memory Storage

VoltAgent's core package (@voltagent/core) includes InMemoryStorageAdapter, a simple storage adapter (for the Memory class) that stores conversation history in application memory.

Overview​

  • Use Case: Development, testing, demos, or any scenario where persistent memory across application restarts is not required.
  • Pros: Zero external dependencies, extremely fast, easy to use.
  • Cons: All stored data (conversation history, agent state) is lost when the application stops or restarts.
  • Availability: Included directly in @voltagent/core.

Configuration​

By default, agents use in-memory storage without any configuration. To customize (e.g., storage limits), configure it explicitly.

import { Agent, Memory, InMemoryStorageAdapter } from "@voltagent/core";
import { openai } from "@ai-sdk/openai";

// Optional: Configure in-memory storage explicitly
const memory = new Memory({
  storage: new InMemoryStorageAdapter({ storageLimit: 100 }),
});

const agent = new Agent({
  name: "Ephemeral Agent",
  instructions: "An agent using in-memory storage (history resets on restart).",
  model: openai("gpt-4o"),
  memory, // Optional; default is also in-memory
});

// Interactions with this agent will use the in-memory store.
// await agent.generateText("Remember this info.", { userId: "user1", conversationId: "conv1" });
// // If the app restarts here, the above message is lost.
// await agent.generateText("Do you remember?", { userId: "user1", conversationId: "conv1" });

Configuration Options (InMemoryStorageAdapter):

  • storageLimit (number, optional): The maximum number of messages to retain per unique userId/conversationId. Oldest messages are pruned when exceeded. Defaults to 100.

Working Memory​

InMemoryStorageAdapter implements working memory storage for both conversation and user scopes using in‑process metadata fields. Enable it via Memory({ workingMemory: { enabled: true, ... } }). See: Working Memory.

Semantic Search (Embeddings + Vectors)​

The in-memory storage can be combined with AiSdkEmbeddingAdapter and InMemoryVectorAdapter to enable semantic retrieval in development:

import { Memory, AiSdkEmbeddingAdapter, InMemoryVectorAdapter } from "@voltagent/core";
import { InMemoryStorageAdapter } from "@voltagent/core";
import { openai } from "@ai-sdk/openai";

const memory = new Memory({
  storage: new InMemoryStorageAdapter({ storageLimit: 100 }),
  embedding: new AiSdkEmbeddingAdapter(openai.embedding("text-embedding-3-small")),
  vector: new InMemoryVectorAdapter(),
});

When to Use​

  • Development & Testing: Quickly test agent logic without setting up a database.
  • Stateless Use Cases: When conversation history is not needed between sessions or application runs.
  • Demos & Examples: Simple setup for showcasing agent capabilities.
  • Caching Layers: Could potentially be used as a short-term cache in more complex memory strategies (though not its primary design).

Avoid using InMemoryStorage in production environments where conversation history needs to be persistent.

Table of Contents