AgentMemo solves the blank-slate problem in multi-agent systems. By default, every AI agent starts each session with no awareness of what other agents have done or learned. AgentMemo adds a shared memory pool through one MCP server — Agent A stores a finding, Agent B can retrieve it next session. Memory is indexed for semantic search, so agents query by meaning rather than exact key. Built-in conflict detection flags contradictory writes before they corrupt the shared state. Compatible with any MCP-supporting stack, includes a free tier, and installs via a single npm command in under a minute. Useful for developers building multi-agent pipelines where context persistence between agents is a hard requirement.