Shared memory for CrewAI teams
Give your CrewAI agents a shared knowledge base they can all read from and write to. Decisions, context, and learned knowledge flow across your entire crew.
Why This Works
CrewAI excels at orchestrating teams of specialized agents. But without shared long-term memory, each agent starts cold. HyperMemory provides a shared hypergraph that all agents in a crew can access — the researcher stores findings, the analyst recalls them, and the writer uses the full context. Knowledge compounds instead of being siloed.
Quick Setup
Configure HyperMemory as a tool in your CrewAI setup:
from crewai import Agent, Task, Crew
from crewai_tools import MCPServerAdapter
# Connect HyperMemory as an MCP tool
hypermemory = MCPServerAdapter(
server_url="https://api.hypermemory.io/mcp",
headers={"Authorization": "Bearer YOUR_API_KEY"}
)
researcher = Agent(
role="Research Analyst",
tools=[hypermemory],
goal="Research topics and store findings in shared memory"
)
writer = Agent(
role="Content Writer",
tools=[hypermemory],
goal="Recall research findings and produce reports"
)
crew = Crew(agents=[researcher, writer], tasks=[...])Use Case
A marketing team runs a CrewAI crew with a researcher, strategist, and copywriter. The researcher stores competitive analysis and market data in HyperMemory. The strategist recalls this data and stores positioning decisions. The copywriter pulls from both to produce on-brand content. Every run builds on the last — the crew gets smarter over time.