LibreAI / Agents IA / LongMemory
// agents · CaviraOSS/LongMemory

LongMemory

LongMemory est un projet de agents IA open source, créé par CaviraOSS (CaviraOSS/LongMemory sur GitHub). Il compte 4 523 étoiles sur GitHub et 506 forks, il est publié sous la licence Apache-2.0 et a été mis à jour pour la dernière fois sur GitHub le 20 septembre 2026.

LongMemory: local-first memory for AI agents with temporal truth, provenance, and hybrid recall

★ 4,5 k étoiles⑂ 506 forksApache-2.0mis à jour sept. 2026
▦ visuel — le README de ce projet ne contient aucune capture intégrable,
voici donc un emplacement réservé. Le guide ci-dessous est complet malgré tout.
// détails du projet

Détails du projet

Étoiles GitHub
4 523
Forks
506
Licence
Apache-2.0
Dernière mise à jour sur GitHub
20 septembre 2026
Catégorie
Agents IA
Type de fiche
Guide d'installation complet
// fonctionnement

Comment ça marche

  1. Choose your surface

    Use the npm library in-process, the global CLI, the Docker/HTTP server, or the MCP server for agent clients.

  2. Install

    Install the package from npm (or the Python SDK from PyPI).

  3. Initialize memory

    Create a memory instance or run the CLI init to create a local database.

  4. Ingest and recall

    Ingest a first fact or conversation, then recall it with the mode that fits your question.

// fonctionnalités

Pourquoi on l'adore

Temporal truth

Records keep both recorded time and valid time, so facts can change without rewriting history.

Immutable content with provenance

Source, author, confidence, and timestamps travel with memory items; nothing is silently lost.

Four recall modes

Strict, historical, associative, and world-grounded recall over BM25, vector, and graph signals.

Governance and isolation

Tenant, user, project, session, and role scopes, with encryption at rest and a tamper-evident audit trail (per the README).

Many ways to run

In-process library, CLI, HTTP server on port 7331, Docker image, MCP server, dashboard, and VS Code extension.

Flexible embeddings

OpenAI-compatible endpoints, Google Gemini, Amazon Bedrock, Ollama, or a built-in local embedder.

Advertisement
// lancez-le

Lancez-le vous-même

  1. Clone the repository

    Clone the project and enter its directory for the source/server path.

    git clone https://github.com/CaviraOSS/LongMemory.git
    cd LongMemory
  2. Install as a library or CLI

    Add it to a Node project, or install the CLI globally.

    npm install longmemory
    # or: npm install --global longmemory
  3. Initialize and recall

    Initialize a database and run your first recall from the CLI.

    longmemory init
    longmemory recall "current project priorities" --mode associative
// cas d'usage

Ce que vous pouvez créer

  • Give a support agent memory that respects when facts were true
  • Let a coding agent recall project decisions across sessions with provenance
  • Run auditable, scope-isolated memory for multi-tenant agent products
  • Connect Claude or another MCP client to a persistent local memory store
Advertisement
// faq

Questions fréquentes

Who maintains LongMemory?

CaviraOSS publishes LongMemory (server images under ghcr.io/caviraoss); it is an open-source project.

Is LongMemory free?

Yes, Apache-2.0 licensed and local-first — your data stays in a local SQLite database unless you choose otherwise.

Does it need a paid API key?

Not necessarily: local and Ollama embedding options avoid hosted APIs; choosing OpenAI, Gemini, or Bedrock embeddings uses your own keys.

How does it compare to a plain vector database?

A vector store returns similar chunks. LongMemory adds typed records, temporal validity, provenance, governance scopes, and recall modes that combine keyword, vector, and graph signals.

How do I get started?

Run `npm install longmemory` and use `createMemory()`, or install the CLI globally and run `longmemory init` then `longmemory recall`.

// continuez l'exploration

Outils similaires

Agents IA

agent-memory

Long-term memory runtime for AI agents — plain Markdown as truth, local ranked retrieval, no API key

★ 2,6 k étoilesMIT
Voir le guide →
Agents IA

taosmd

Local-first AI memory that runs offline on just 8 GB of RAM.

★ 79 étoilesAGPL-3.0
Voir le guide →
Agents IA

memory-lancedb-pro

Memory-LanceDB-Pro gives OpenClaw agents hybrid search, smart extraction, and intelligent forgetting

★ 4,5 k étoilesMIT
Voir le guide →
Agents IA

basic-memory

Basic Memory: local-first MCP memory so you and your AI tools build a shared knowledge graph in Markdown

★ 4,1 k étoilesAGPL-3.0
Voir le guide →
Agents IA

cognee

Open-source AI memory that gives agents persistent long-term recall.

★ 31,6 k étoilesApache-2.0
Voir le guide →
Agents IA

EverOS

One portable, local-first memory layer for every AI agent - Markdown-native, user-owned and self-evolving.

★ 13,4 k étoilesApache-2.0
Voir le guide →
// alternatives

Alternatives en Agents IA

Les autres projets de la catégorie Agents IA avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si LongMemory ne vous convient pas.

  • openclaw★ 391,6 k

    A personal AI that really does things — on any OS, with your data staying yours.

  • superpowers★ 294,9 k

    An agentic skills framework & software development methodology that works.

  • ECC★ 275,2 k

    Agent harness optimization system with skills, memory and security for Claude Code, Codex and more

  • hermes-agent★ 252,0 k

    The agent that grows with you — a Python framework for adaptive AI agents.

  • deepseek-harness★ 245,5 k

    DeepSeek's plugin-based agent harness where everything is a plugin.

// plus en agents

Plus en Agents IA

Projets de la catégorie Agents IA proches dans le catalogue, les plus proches en étoiles de LongMemory — continuez à explorer de part et d'autre.

  • council-of-high-intelligence★ 4,6 k

    Convene 18 legendary thinkers as an AI council that debates your hardest questions — inside Claude Code

  • m_flow★ 4,5 k

    M-flow is a bio-inspired cognitive memory engine for AI agents — a graph alternative to flat RAG

  • PyRIT★ 4,6 k

    Microsoft's open framework for red-teaming generative AI systems.

  • logfire★ 4,5 k

    Pydantic Logfire: Python and AI observability on OpenTelemetry — SQL-queryable traces, metrics, and logs

  • koog★ 4,6 k

    Koog is a JVM (Java and Kotlin) framework for building predictable, fault-tolerant and enterprise-ready AI agents across all platforms – from backend service…

  • embabel-agent★ 4,5 k

    Agent framework for the JVM. Pronounced Em-BAY-bel /ɛmˈbeɪbəl/

Obtenir LongMemory

Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.