Akasha MCP server — use your second brain as context in Claude, Cursor, and any other MCP-compatible LLM tool.
Connect Akasha to Claude Desktop and every conversation automatically has your full context — your projects, decisions, recent activity, and knowledge graph.
- A running Akasha instance
- An Akasha API key (generated in your workspace settings)
Add this to your Claude Desktop config:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"akasha": {
"command": "npx",
"args": ["@akasha-os/mcp"],
"env": {
"AKASHA_URL": "http://localhost:3000",
"AKASHA_API_KEY": "your-api-key"
}
}
}
}Restart Claude Desktop. You should see Akasha appear in the tools panel.
| Tool | Description |
|---|---|
search_graph |
Search your knowledge graph by keyword or concept |
get_project_context |
Full context for a project — activity, documents, decisions |
get_today_context |
Today's biometrics, tasks, open loops, and agent insights |
get_recent_activity |
Your activity from the last N hours across all sources |
get_insights |
Unread agent insights and nudges |
Once connected, ask Claude:
What should I focus on today?
Catch me up on the Akasha project
What do I know about vector databases?
Who have I talked to most about this topic?
Claude will query your Akasha graph and answer from your actual context — not from thin air.
The MCP server only exposes data from your own Akasha instance. Private nodes are never included in MCP responses. You control what is accessible via workspace settings.