Blaster

Python MCP Server for collaborative content structuring. Part of the Teletran1 ecosystem.

Overview

Blaster transforms sparse user input into full, engaging articles through 5 specialized MCP tools. It works collaboratively with Claude Desktop or Claude Code - you provide ideas, Blaster provides structure.

Features

Tool	Purpose
generate_outline	Topic → structured article outline with sections and word targets
expand_section	Brief description → full content with examples and transitions
check_completeness	Flag thin sections, identify missing elements, completeness score
score_engagement	Hook/flow/takeaway scoring (0-100) with improvement suggestions
format_for_platform	Optimize for LinkedIn, Medium, Dev.to, Twitter, or generic

Installation

Quick Start (Recommended)

npx blaster-mcp

This automatically:

• Detects Python 3.11+
• Creates virtual environment
• Installs dependencies
• Runs the server

Global Installation

npm install -g blaster-mcp

Claude Desktop Configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "blaster": {
      "command": "npx",
      "args": ["blaster-mcp"]
    }
  }
}

Or if globally installed:

{
  "mcpServers": {
    "blaster": {
      "command": "blaster-mcp"
    }
  }
}

Claude Code Configuration

Add to your MCP settings:

{
  "mcpServers": {
    "blaster": {
      "command": "npx",
      "args": ["blaster-mcp"]
    }
  }
}

Manual Development Setup

# Clone the repository
git clone https://github.com/MatthewSnow2/blaster.git
cd blaster

# Install (runs setup automatically)
npm install

# Or run setup manually
npm run postinstall

Requirements

• Node.js 18+ (for npm tooling)
• Python 3.11+ (automatically detected)

Usage Examples

Generate an Article Outline

Use generate_outline with:
- topic: "Building Effective AI Agents"
- key_points: ["Architecture patterns", "Error handling", "Testing strategies"]
- audience: "intermediate"
- content_type: "tutorial"
- target_length: 2000

Expand a Section

Use expand_section with:
- title: "Error Handling Best Practices"
- description: "Cover retry logic, graceful degradation, and user feedback"
- tone: "professional"
- target_words: 400

Check Content Completeness

Use check_completeness with:
- content: [your full article text]
- outline: [optional - the original outline for comparison]

Score Engagement

Use score_engagement with:
- content: [your article text]
- platform: "linkedin"
- title: "10 Lessons from Building AI Agents"

Format for Platform

Use format_for_platform with:
- content: [your article text]
- platform: "medium"
- title: "Your Article Title"
- include_hashtags: true

Development

Running Tests

source .venv/bin/activate
pytest tests/ -v

Test Coverage

pytest tests/ -v --cov=blaster --cov-report=term-missing

Code Quality

ruff check src/
mypy src/

Architecture

blaster/
├── src/blaster/
│   ├── server.py           # MCP server entry point
│   ├── models/schemas.py   # Pydantic input/output schemas
│   ├── tools/              # 5 MCP tools
│   ├── errors/             # Error handling
│   └── utils/              # Logging utilities
└── tests/                  # 67 tests

Teletran1 Ecosystem

Teletran1 (Orchestrator)
├── Perceptor (memory/context)
├── Sky-lynx (meta-learning)
├── Idea Triage (conversation intelligence)
├── Blaster (content structuring)  ← THIS
└── Grimlock (task execution)

Built With

• Python 3.11+
• MCP SDK - Model Context Protocol
• Pydantic 2.0 - Data validation
• pytest - Testing framework

Development Method

Built using ralph-loop autonomous development in a single iteration.

License

MIT

Author

Matthew Snow - Me, Myself Plus AI LLC