One memory for your MCP clients, and every tool you will use
Cursor, Claude Code, custom agents, or internal tools: point them at Hypermemory and share one graph. Switch clients without exporting chats or re-indexing files.
Why This Works
Hypermemory is accessed via MCP and the CLI, the same memory graph either way. MCP clients use the connector URL with OAuth or API keys; agents without MCP use the CLI and skill. Every path exposes the same 12 hm_ tools on your tenant.
Quick Setup
MCP URL: https://api.hypermemory.io/mcp
- Copy the MCP URL below and add it as a custom connector in your client.
- Connect and log in with your account info (or use an API key and graph ID from the app).
- Add the Hypermemory skill or enable the hm_ tools in your agent.
- Start storing and recalling, one graph across every client you connect.
Available Tools
| Tool | Description |
|---|---|
| hm_find_related | Traverse connected knowledge from a starting node |
| hm_forget | Delete a node with optional cascade |
| hm_get_overview | Graph stats, top nodes, and type distribution |
| hm_ingest | Decompose dense text into graph entities in one pass |
| hm_list_files | Query uploaded files by type or name |
| hm_list_orphans | List disconnected nodes for cleanup |
| hm_recall | Hybrid search across your graph (BM25, vector, edges) |
| hm_store | Create a memory node; background enrichment runs automatically |
| hm_timeline | Temporal recall by period or date range |
| hm_timeline_write | Write an explicit diary-style timeline entry |
| hm_update | Update a node; re-embeds when the description changes |
| hm_upload_file | Upload a file into memory (summary + graph node) |
Use Case
Engineering uses Cursor with Hypermemory for codebase decisions; product uses Claude for roadmap notes; ops uses an internal MCP agent for incidents. All three write to one graph, you switch tools, not memory silos.