Skip to main content
Glama
README.md
# Tom Osborne's Second Brain — MCP server

> Talk to Tom Osborne's voice-cloned AI agent from inside Claude Desktop, Claude Code, Cursor, Cline, or any [Model Context Protocol](https://modelcontextprotocol.io)–compatible client.

Built on Tom's 100-question interview corpus, 15 years of operator memory, and **21,000+ curated marketing knowledge-base articles** across 17 verticals (paid, content, SEO, sales, leadership, PR, ads, design, HR, finance, legal, local, marketing fundamentals, business strategy, creator, community, entrepreneurship).

## What it does

Adds one tool to your MCP client:

```
ask_tom(question: string) → Tom's answer in his voice, with KB chunks retrieved
```

When called, it POSTs your question to `https://tomosborne.me/api/cmo/chat` along with your configured email, and returns Tom's reply. The same pipeline that powers `tomosborne.me/cmo` — same retrieval, same security gate, same voice.

## Installation

### 1. Install via npm

```bash
npm install -g @tom-osborne/second-brain-mcp
```

(Or run via `npx @tom-osborne/second-brain-mcp` if you prefer not to install globally.)

### 2. Add to your MCP client config

#### Claude Desktop

Edit `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):

```json
{
  "mcpServers": {
    "second-brain": {
      "command": "npx",
      "args": ["-y", "@tom-osborne/second-brain-mcp"],
      "env": {
        "MCP_TOM_EMAIL": "you@yourcompany.com"
      }
    }
  }
}
```

Restart Claude Desktop. The `ask_tom` tool will appear in the MCP tools panel.

#### Claude Code

Add to your project's `.mcp.json`:

```json
{
  "mcpServers": {
    "second-brain": {
      "command": "npx",
      "args": ["-y", "@tom-osborne/second-brain-mcp"],
      "env": {
        "MCP_TOM_EMAIL": "you@yourcompany.com"
      }
    }
  }
}
```

Or via the CLI:

```bash
claude mcp add second-brain -e MCP_TOM_EMAIL=you@yourcompany.com -- npx -y @tom-osborne/second-brain-mcp
```

#### Cursor / Cline / other MCP clients

Use the same shape — point at `npx @tom-osborne/second-brain-mcp` with `MCP_TOM_EMAIL` in env.

## Usage

Once installed, in any MCP-aware chat:

> *Use ask_tom: how do I get to my first 100 customers as a DTC brand?*

The tool returns Tom's answer in his voice, drawing on his interview corpus + the 21k-article KB.

## Quotas

- **Free tier:** 5 questions per day per email (rate-limited at the API)
- **Coffee unlock:** +20 questions/day + faster replies + voice replies — buy at https://tomosborne.me/cmo
- **Loyalty:** every 6 coffees = 24h/7d/30d VIP unlock (unlimited premium)

If you hit the cap, the tool returns a clear "go upgrade" message rather than failing silently.

## Environment variables

| Variable | Required | Default | Purpose |
|---|---|---|---|
| `MCP_TOM_EMAIL` | **Yes** | — | Your email — used for daily quota tracking. Must be a real email. |
| `MCP_TOM_API_KEY` | No | — | Optional API key for higher-tier access (gate is server-side; ignored if not configured) |
| `TOM_API_BASE` | No | `https://tomosborne.me` | Override the API base (only useful for dev/testing) |

## Privacy

- Your questions are sent to `tomosborne.me/api/cmo/chat` over HTTPS
- Email is stored against your daily quota counter; never shared, never sold
- The Pages Function applies the same security filter set as the public `/cmo` chat (no wallet addresses, API keys, or PII in responses)
- Source: [tomosborne.me](https://tomosborne.me) is open-source-ish in spirit — DM Tom if you want to peek under the hood

## Local development

```bash
git clone https://github.com/DegenDoes/second-brain-mcp.git
cd second-brain-mcp
npm install
MCP_TOM_EMAIL=test@you.com npm run dev   # tsx in watch mode
```

Test it via the MCP Inspector:

```bash
npx @modelcontextprotocol/inspector npx -y @tom-osborne/second-brain-mcp
```

## License

MIT. Use it however you like. If you build something cool with it, [tell Tom](mailto:contact@tomosborne.me).

TDQS

A4.2/5.0

Scored across 1 tool

Disambiguation5/5

Only one tool exists, so no ambiguity between tools. The tool's purpose is clearly defined as asking Tom's brain a question.

Naming Consistency5/5

With a single tool named 'ask_tom' (verb_noun), naming is trivially consistent and follows a clear pattern.

Tool Count3/5

One tool for a broad domain (marketing, growth, etc.) feels thin, but it aligns with the server's focused purpose of Q&A. Still borderline low.

Completeness4/5

The single tool covers the primary use case of asking questions and returning answers. Minor gaps like advanced filtering or knowledge management are not expected for this specific server.