Skip to main content

Share Memory Across Editors

One memory database. Every editor. Configure OMEGA once and your agents in Claude Code, Cursor, and Windsurf all share the same persistent memory.

Store a decision in Claude Code, recall it in Cursor. No sync, no duplication, no context loss.

Step-by-Step Guide

1

Install OMEGA

Install the OMEGA memory package. This is the same step regardless of which editors you use.

pip install omega-memory

All editors will connect to the same OMEGA installation and share the same memory database.

2

Configure Claude Code

Add OMEGA to your Claude Code MCP settings. Claude Code reads MCP config from its settings file.

// ~/.claude/claude_desktop_config.json
{
  "mcpServers": {
    "omega-memory": {
      "command": "python3.11",
      "args": ["-m", "omega", "serve"]
    }
  }
}

Claude Code also supports project-level MCP config in .mcp.json at the project root.

3

Configure Cursor

Add the same OMEGA MCP configuration to Cursor.

// ~/.cursor/mcp.json
{
  "mcpServers": {
    "omega-memory": {
      "command": "python3.11",
      "args": ["-m", "omega", "serve"]
    }
  }
}

The command is identical. Both editors launch the same OMEGA server and access the same database.

4

Configure Windsurf (optional)

If you also use Windsurf, add the same configuration there.

// ~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "omega-memory": {
      "command": "python3.11",
      "args": ["-m", "omega", "serve"]
    }
  }
}

Same server, same database, same memory. The config file location is the only difference.

5

Verify shared memory

Store a memory in one editor and retrieve it in another to confirm sharing works.

// In Claude Code:
"Store this: the project uses PostgreSQL 16 for the backend database."

// In Cursor (new session):
"What database does this project use?"
// Agent should recall: PostgreSQL 16

Memories stored in any editor are immediately available in all others. No sync delay.

Why Share Memory?

Unified context

A decision stored in Claude Code is available when you switch to Cursor. No context loss when switching editors.

No duplication

One memory database means no conflicting memories or sync issues between editors.

Consistent preferences

Store coding style preferences once and every editor's agent follows them.

Multi-agent coordination

OMEGA Pro's coordination features let agents in different editors collaborate without conflicts.

Single source of truth

Architecture decisions, API contracts, and project context live in one searchable place.

Zero overhead

Each editor launches its own OMEGA process, but they all read/write the same SQLite database.

Frequently Asked

Can two editors access memory at the same time?

Yes. OMEGA uses SQLite with WAL (Write-Ahead Logging) mode, which supports concurrent readers and writers. Multiple editor instances can safely read and write memories simultaneously without data corruption.

Will memories conflict if I use Claude Code and Cursor on the same project?

No. OMEGA stores memories in a single database with timestamps and source metadata. Both editors write to the same store. OMEGA's contradiction detection will flag if conflicting information is stored, regardless of which editor stored it.

Does each editor run its own OMEGA server?

Yes. Each editor launches its own OMEGA process via the MCP protocol. These processes share the same SQLite database file. OMEGA's embedded daemon mode ensures embedding computations are shared efficiently.

Can I use project-specific memory instead of global memory?

Yes. OMEGA supports project-scoped memory via entity_id and tags. You can also use project-level MCP config files (.mcp.json) to scope memory to specific projects. The default behavior shares memory globally.

One memory, every editor

Stop losing context when you switch tools. Share memory everywhere.