Skip to main content
Claude Mem MCP tools work with Claude Desktop on macOS and Windows.
Claude Desktop can access your Claude Mem memory database through MCP tools. Once configured, you can search past sessions, decisions, and observations directly from any Claude Desktop conversation.

​
Prerequisites

Before configuring MCP tools, ensure:
  1. Claude Mem is installed and the worker service is running.
  2. The MCP server file is present at ~/.claude/plugins/marketplaces/thedotmack/plugin/scripts/mcp-server.cjs.

​
Verify the worker is running

curl http://localhost:37777/api/health
# Expected response: {"status":"ok"}

​
Installation

1

Edit the Claude Desktop configuration file

Open the configuration file for your platform:
~/Library/Application Support/Claude/claude_desktop_config.json
Add the mcp-search server entry:
{
  "mcpServers": {
    "mcp-search": {
      "command": "node",
      "args": [
        "/Users/YOUR_USERNAME/.claude/plugins/marketplaces/thedotmack/plugin/scripts/mcp-server.cjs"
      ]
    }
  }
}
Replace YOUR_USERNAME with your actual system username.
2

Restart Claude Desktop

Close and reopen Claude Desktop for the MCP server configuration to take effect.

​
Searching memory from Claude Desktop

Once installed, you can ask Claude Desktop natural-language questions about your past work: 
What did we do last session?
Did we fix this bug before?
How did we implement authentication?
What decisions did we make about the API?
Show me changes to worker-service.ts
Claude will use the MCP tools to search your memory database and return relevant observations from past Claude Code sessions.

​
Available MCP tools

The MCP server provides three tools that follow the same 3-layer workflow used in Claude Code:
ToolDescription
searchSearch the memory index. Returns a compact table with IDs for filtering.
timelineGet chronological context around a query or observation ID.
get_observationsFetch full observation details by ID — use after filtering with search or timeline.

​
Token-efficient workflow

  1. Search — get index with IDs (~50–100 tokens per result)
  2. Timeline — get context around interesting results
  3. Get observations — fetch full details only for the IDs you need
This 3-layer approach provides approximately 10x token savings compared to fetching full observation details upfront.

​
Troubleshooting

  1. Verify the worker is running: curl http://localhost:37777/api/health
  2. Check that the MCP server file path in the configuration matches the actual file location
  3. Check Claude Desktop logs for errors:
tail -f ~/Library/Logs/Claude/claude.log
  1. Ensure Claude Mem has recorded at least one Claude Code session
  2. Verify the database exists: ls ~/.claude-mem/claude-mem.db
  3. Test the worker API directly: curl "http://localhost:37777/api/search?query=test"
  4. Open the viewer at http://localhost:37777 to confirm data is present
Claude Desktop must be fully restarted (not just the window) after editing claude_desktop_config.json. Quit from the menu bar or system tray, then relaunch.

Search tools reference

Full MCP tool documentation with parameter details

Platform integration

Build custom integrations with the worker HTTP API

Build docs developers (and LLMs) love

Get started for freeTalk to us