Integration Guides

CrewAI

Use HyperMemory as persistent memory for CrewAI agents

CrewAI

CrewAI agents can use HyperMemory through MCP. Connect to the hosted MCP server and call tools via an MCP client in your CrewAI tool wrappers.

Setup

Set your API key:

export HYPERMEMORY_API_KEY=hm_YOUR_KEY_HERE

Generate API keys at app.hypermemory.io under API Keys in the main navigation.

Create tool wrappers that call the HyperMemory MCP endpoint:

import os
import httpx

MCP_URL = "https://api.hypermemory.io/mcp"
API_KEY = os.environ["HYPERMEMORY_API_KEY"]

def call_mcp(tool_name: str, arguments: dict) -> dict:
    response = httpx.post(
        MCP_URL,
        json={
            "jsonrpc": "2.0",
            "id": 1,
            "method": "tools/call",
            "params": {"name": tool_name, "arguments": arguments},
        },
        headers={
            "Authorization": f"Bearer {API_KEY}",
            "Content-Type": "application/json",
        },
        timeout=30.0,
    )
    return response.json().get("result", {})

Then wrap call_mcp as a CrewAI BaseTool and attach it to your agents.

Key tools for agents

ToolPurpose
hm_storePersist a fact, decision, or preference
hm_recallHybrid search (BM25 + vector + regex)
hm_find_relatedTraverse from a node
hm_get_overviewGraph stats and top nodes
hm_ingestDecompose dense text into graph entities
hm_forgetRemove outdated info

See API Reference for all 12 tools.