caura-ai/caura

Caura (formerly MemClaw) — governed shared memory for AI agent fleets. Multi-agent, multi-tenant, MCP-native. Trust tiers, keystone policies, audit trails, knowledge graph, self-improving retrieval. Apache 2.0.

What it solves

Caura provides a governed, shared memory layer for fleets of AI agents. It prevents agents from repeating mistakes by allowing them to store and recall learned information across different agents, teams, and vendors, while maintaining strict control over who can access specific memories.

How it works

Agents write plain text to Caura, which the system then enriches using a single-pass LLM process to classify memory types, generate summaries, and extract entities. It uses a hybrid search approach—combining semantic similarity (pgvector), full-text keyword matching, and a knowledge graph—to retrieve relevant information. The system also includes a contradiction detection mechanism that automatically supersedes conflicting memories.

Who it’s for

Developers building multi-agent systems or "agent fleets" that require shared knowledge, multi-tenant isolation, and auditable memory governance for production environments.

Highlights

  • Fleet Governance: Implements tenant isolation, visibility scopes (agent, team, or organization), and agent trust tiers to control memory access.
  • Crystallization: Automatically merges near-duplicate memories into canonical facts to reduce redundancy.
  • Outcome-based Learning: Reinforces successful actions and generates preventive rules based on agent failures.
  • MCP Native: Built-in support for the Model Context Protocol, allowing easy integration with clients like Claude Desktop and Cursor.
  • Knowledge Graph: Automatically extracts people, organizations, and concepts into a live graph for enhanced retrieval.

Related

  • Project
    mezmo/aura

    AURA is an open‑source SRE‑agent platform that lets you run AI‑driven incident‑investigation assistants inside your own infrastructure. It supports multiple LLM providers, configurable guardrails, OpenTelemetry tracing, and dozens of integrations (Kubernetes, Prometheus, PagerDuty, etc.). Deployable as a CLI, Docker image, Helm chart, or embedded Rust library.

  • Project
    matrixorigin/memoria

    A persistent memory layer for AI agents featuring Git-like version control, allowing for snapshots, branching, and hybrid semantic search to maintain auditable long-term knowledge.

  • Project
    agentic-box/memora

    An MCP memory layer for AI agents that provides persistent collective memory through structured storage, semantic retrieval, and a knowledge graph.

  • Project
    neo4j-labs/agent-memory

    A graph-native memory system for AI agents that uses Neo4j to store conversations, build knowledge graphs, and record reasoning traces.

  • Project
    atomicstrata/atomicmemory

    AtomicMemory is an open‑source, inspection‑ready memory service for LLM agents. It offers correction‑aware updates, model‑agnostic embeddings, and both managed‑cloud and self‑hosted Docker deployments. The project ships a Rust Core server, a TypeScript SDK, a CLI (`am`), adapters for LangChain, LangGraph, Vercel AI, etc., and plugins for various agent hosts. Quick start is a one‑line installer that creates a cloud profile; a `--local` flag runs the Core container on your machine. Memory can be ingested and queried via the CLI or programmatically via the SDK, and agents can be linked through an MCP server. Licensed under Apache 2.0.