pūrmemo MCP Server

Claude knows who you are before you say a word.

pūrmemo gives your AI a persistent memory and identity layer — your role, expertise, active projects, and conversation history — available instantly in every session, across every platform.

Using ChatGPT, Claude.ai, or Gemini in browser? Get the Chrome Extension instead.

What It Does

• Remembers everything — save conversations, decisions, and context; search them later with natural language
• Knows who you are — your role, expertise, tools, and active projects load automatically at session start
• Works everywhere — Claude Code, Claude Desktop, Cursor, Windsurf, Zed, and any MCP-compatible platform

Quick Start

One command:

npx purmemo-mcp@latest init

This opens your browser to sign in, configures the MCP server, and installs hooks + slash commands (/save, /recall, /context). That's it.

Manual Setup (alternative)

If you prefer to configure manually, or you're not using Claude Code:

1. Get your API key from app.purmemo.ai → Settings → API Keys

claude mcp add purmemo -e PURMEMO_API_KEY=your-api-key-here -- npx -y purmemo-mcp

Verify it connected:

claude mcp list
# purmemo: npx -y purmemo-mcp - ✓ Connected

Use pūrmemo's hosted MCP server — no API key setup required, authenticates via OAuth:

1. Open Claude Desktop → Settings → Developer → Edit Config
2. Add this configuration:

{
  "mcpServers": {
    "purmemo": {
      "url": "https://mcp.purmemo.ai/mcp/messages",
      "transport": "streamable-http"
    }
  }
}

1. Restart Claude Desktop
2. You'll be prompted to sign in via OAuth

Edit your config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "purmemo": {
      "command": "npx",
      "args": ["-y", "purmemo-mcp"],
      "env": {
        "PURMEMO_API_KEY": "your-api-key-here"
      }
    }
  }
}

Restart Claude Desktop after saving.

Edit ~/.cursor/mcp.json (macOS) or %USERPROFILE%\.cursor\mcp.json (Windows):

{
  "mcpServers": {
    "purmemo": {
      "command": "npx",
      "args": ["-y", "purmemo-mcp"],
      "env": {
        "PURMEMO_API_KEY": "your-api-key-here"
      }
    }
  }
}

Edit ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "purmemo": {
      "command": "npx",
      "args": ["-y", "purmemo-mcp"],
      "env": {
        "PURMEMO_API_KEY": "your-api-key-here"
      }
    }
  }
}

Add to ~/.config/zed/settings.json under the context_servers key:

{
  "context_servers": {
    "purmemo": {
      "command": {
        "path": "npx",
        "args": ["-y", "purmemo-mcp"],
        "env": {
          "PURMEMO_API_KEY": "your-api-key-here"
        }
      }
    }
  }
}

Start Using

You: "What's the project status?"
Claude: Based on your identity and recent memories:
  You're a founder working on a B2B SaaS product.
  Recent work: pūrmemo (15 sessions), auth refactor (3 sessions)
  Last session: "Fixed JWT refresh token rotation"

What You Get

Resources (attach to any conversation via the + button)

Example — attach memory://me at session start:

You are working with:
**Chris** — Founder, B2B SaaS
Expertise: product, fullstack, ai
Tools: cursor, claude, supabase
Style: systems thinker

Recent work:
- pūrmemo (15 recent sessions)
- auth-refactor (4 recent sessions)

Working on: MCP Resources + Prompts feature

No re-explaining who you are. No repeating your stack. Just continue.

Prompts (conversation starters in the + menu)

Tools

Core

Workflows

Community

Tasks & Investigation

get_user_context in action:

You: "What have I been working on?"
Claude: [calls get_user_context]

Your profile: Founder · B2B SaaS · fullstack/ai/product
Active projects:
  • pūrmemo — "MCP server resources and prompts" (15 sessions)
  • auth-refactor — "JWT refresh token fix" (4 sessions)
Working on: MCP Resources + Prompts feature

Identity Layer

pūrmemo maintains a cognitive fingerprint — a persistent profile of who you are that loads automatically into every session.

Set it once at app.purmemo.ai → Settings → Identity:

• Role — founder, engineer, designer, researcher, ...
• Domain — your primary field (B2B SaaS, ML research, design systems, ...)
• Expertise — your key skills (product, fullstack, ai, ...)
• Tools — what you work with (cursor, claude, supabase, ...)
• Work style — how you think (systems thinker, iterative builder, ...)
• Working on — your current focus, updated per session

Once set, every new session inherits this context. Claude already knows your background, your stack, and what you were doing last time — without you having to explain it.

Slash Commands (Claude Code)

After installing the slash commands (see Claude Code setup above):

The /context command is especially useful at the start of a session: it calls get_user_context and surfaces your identity and recent work so Claude already knows where you left off.

Living Document Pattern

Same title = update, not duplicate. Build on memory over time:

You: "Save this as auth-refactor"
Claude: ✅ Saved — "auth-refactor" (new)

[... continue working across multiple sessions ...]

You: "Save as auth-refactor"
Claude: ✅ Updated — "auth-refactor" (3 updates, not 3 copies)

Long conversations? Auto-chunked at 100K+ characters and reassembled on recall.

Pricing

Links

• Dashboard — View and manage memories
• Chrome Extension — For ChatGPT, Claude.ai, Gemini
• Documentation
• Support

Privacy

Your data is encrypted in transit (HTTPS) and at rest. It is never shared with third parties and is accessible only to you via your API key.

See our Privacy Policy for details.

License

This MCP client package is MIT licensed — you can use, fork, and modify it freely.

The pūrmemo platform, API, and backend are proprietary and closed-source. MIT applies only to the connector code in this repository, not to the service it connects to.