LongMemory

LongMemory is a local-first memory engine providing temporal, governed data storage for LLM applications and autonomous agents.

LongMemory functions as a cognitive memory engine that integrates with LLM applications to store and retrieve data with context. It supports features like temporal truth, where recorded time and valid time are managed separately, and immutable content that preserves provenance. The engine uses SQLite for local-first storage and allows for associative, historical, and world-grounded recall. Developers use it to manage project memory, skills, and chat histories while enforcing governance and lifecycle policies. It integrates with various agent hosts, automation tools, and Python frameworks to provide memory to otherwise stateless AI agents.

LongMemory at a glance

Best for: Software engineer
Runs on: Web, CLI, VS Code, Docker, API, MCP
Pricing: Free
Released: October 2025
Category: AI agent tool
Website: github.com/CaviraOSS/LongMemory

Key features

Durable local-first storage using SQLite
Temporal reasoning with event ordering and supersession
Immutable content with provenance and hashes
Governed memory for projects, tenants, and roles
Multilingual memory with cross-language embeddings
Explainable evidence selection and token-bounded context
Connectors for local files, repositories, and web content

Use it for

Storing persistent chat memory for autonomous agents
Managing project context like architecture and task decisions
Retrieving grounded evidence for AI-generated answers
Importing and governing coding agent session logs

Alternatives to LongMemory

Supermemory — Integrates long-term memory into AI coding agents with a single prompt.
AgentMemory — It runs locally with zero external databases and achieves high retrieval recall.
Adaptive Agent Memory Research — It decides what to remember based on measured future utility rather than query similarity.
ReMe — It enables local-first, file-native long-term memory sharing across multiple AI agents.
Tigerless Agent Memory — Maintains agent history in portable, greppable Markdown files without API keys.
Mnemosyne — It provides sub-millisecond memory access with no external service costs.
SimpleMem — Provides semantic lossless compression to reduce token consumption and maintain long-term context.
Memori — Reduces LLM token usage and spend by over 97% with targeted recall.
SuperLocalMemory — It refuses to guess or bluff when information is missing from its memory store.
claude-mem — It shares persistent, searchable agent memory across sessions and devices.
memU — It gives every connected agent a shared and inspectable memory layer.
Agentmemoryrepo — Enables agents to navigate memories like a wiki and share knowledge across sessions.

All LongMemory alternatives