# RPCS-1 Agent Tuner MCP server

> Optimizes AI agent configurations and communication profiles by diagnosing environmental entropy, predictability, and task stakes.

- Repository: https://github.com/travisbergen2/rpcs1-sdk
- Page: https://hostedmcp.org/servers/travisbergen2/second-brain
- Transport: stdio
- Stars: 0
- License: MIT
- Latest version: 0.1.0

## Install

### Claude Code

```sh
claude mcp add second-brain -- npx -y @travisbergen2/second-brain-mcp
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "second-brain": {
      "command": "npx",
      "args": [
        "-y",
        "@travisbergen2/second-brain-mcp"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "second-brain": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@travisbergen2/second-brain-mcp"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"second-brain","type":"stdio","command":"npx","args":["-y","@travisbergen2/second-brain-mcp"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "second-brain": {
      "command": "npx",
      "args": [
        "-y",
        "@travisbergen2/second-brain-mcp"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add second-brain -- npx -y @travisbergen2/second-brain-mcp
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.second-brain]
command = "npx"
args = ["-y", "@travisbergen2/second-brain-mcp"]
```

### Gemini CLI

```sh
gemini mcp add -s user second-brain npx -- -y @travisbergen2/second-brain-mcp
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "second-brain": {
      "command": "npx",
      "args": [
        "-y",
        "@travisbergen2/second-brain-mcp"
      ]
    }
  }
}
```

## FAQ

### What can the agent tuner do?

It diagnoses AI agents against environmental entropy, predictability, stakes, context horizon, and commitment style to provide runtime settings and validation tests.

### Does the server make model calls?

No, the SendRight hand-off path is designed to work without making model calls or requiring API keys, ensuring zero data custody.

### How does the recommendation engine work?

The SDK implements the Matching Principle, which relates stable receivers in an environment with entropy to specific attention windows.
