OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

Longmemory

An open-source TypeScript memory engine and self-hosted service for LLM applications and AI agents. It provides local-first persistent storage, temporal reasoning, governed recall and multiple integration surfaces.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Longmemory — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Longmemory when an LLM application needs more than basic vector retrieval: for example, when facts can change over time, superseded information matters, access scope must be enforced, or recall needs explainable evidence and token limits.

Who it is for

  • Developers building AI agents that must retain information across sessions
  • Teams creating self-hosted LLM workflows with governed project or user memory
  • TypeScript applications that need SQLite-backed memory
  • Python applications that need to access a self-hosted memory service

Documented uses

  • Store durable agent and application memory locally
  • Recall current, historical, associative or externally grounded information
  • Preserve provenance, temporal history and immutable content
  • Apply scope, permission and lifecycle controls to project memory
  • Expose memory through npm, CLI, HTTP, MCP, dashboard, VS Code or Python
Before you choose

Requirements & limits

The repository is public and uses the Apache-2.0 license. The evidence documents npm and PyPI packages, self-hosted deployment, Docker and Docker Compose options, and a CLI.

  • Choose the npm library for direct TypeScript integration and SQLite-backed local persistence.
  • Choose the self-hosted service when applications need HTTP access or Python SDK integration.
  • Use historical recall when superseded facts must remain available for point-in-time questions.
  • Consider the governance and evidence model when permissions, provenance, contradictions and token-bounded context affect agent behaviour.
  • The documented design is local-first and self-hosted rather than presented as a hosted memory service.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Use LongMemory as an in-memory TypeScript memory store

Install LongMemory, ingest a fact, recall it with strict mode, and close the memory instance.

Before you start

  • A TypeScript application with npm available.
  • The longmemory npm package.

Steps

  1. Install the package with the command below.

    Documented command
    npm install longmemory

  2. Import createMemory from longmemory and create a memory instance.

  3. Ingest a record containing user_id: 'alice' and the text I prefer TypeScript for backend services.

  4. Call memory.recall with the query What language does Alice prefer? and mode: 'strict', then close the instance with memory.close().],

  5. outcome comment? no

Expected outcome

The application receives a strict recall result for the ingested preference without requiring a service or external database for in-memory use.

Guide limitations

  • This recipe documents in-memory use only; persistence requires the SQLite configuration described separately.
  • The source does not document a specific returned result format or a successful test run.
Documentation 1 ↗
Source-based practical guide

Run LongMemory from source as a local service

Clone, install, build, and start the LongMemory service from its repository source.

Before you start

  • Git, Corepack, pnpm, and the project source repository.
  • The repository’s documented build environment.

Steps

  1. Clone https://github.com/CaviraOSS/LongMemory.git and change into the repository directory.

  2. Enable Corepack with corepack enable.

  3. Install dependencies with pnpm install --frozen-lockfile.

  4. Build the project with pnpm build.

  5. Start the service with pnpm start; the documented default API address is http://127.0.0.1:7331.

Expected outcome

A locally running LongMemory service is started from source and listens on the documented default address.

Guide limitations

  • The source does not document authentication setup for this from-source recipe.
  • The source does not document a client request or a successful health-check result for this sequence.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
30 September 2026
Original website title
CaviraOSS/LongMemory · Official homepage title · retained in its source language
How we review sourcesSuggest a correction