Membase

Decentralized memory layer for AI agents — persistent, verifiable storage for conversations, knowledge bases, and on-chain task coordination. SDK, MCP, or Skill integration.

30-Second Example

pip install git+https://github.com/unibaseio/membase.git

from membase.memory.buffered_memory import BufferedMemory
from membase.memory.message import Message

memory = BufferedMemory(membase_account="default", auto_upload_to_hub=True)
memory.add(Message(name="my-agent", content="Hello!", role="assistant", metadata=""))
# View at https://hub.membase.unibase.com

Why Membase

vs. Centralized DB	vs. Local Storage
Verifiable on-chain	Syncs to Hub, survives restarts
ZK-verified access	Multi-agent shared memory
MCP/Skill — no custom backend	Chain tasks for collaborative rewards

Integration Options

Option	Best for	Setup
SDK	Custom agents, full control	pip install git+https://github.com/unibaseio/membase.git
MCP	Claude Desktop, Cline, etc.	membase-mcp
Skill	BitAgent, skill-based frameworks	Install skill, configure

Next Steps

• Quick Start — Multi-Memory, Knowledge Base, Chain Tasks

• Integration Options

• Architecture

Resources

• Membase GitHub · Memory Hub