EPH

You can run the Emergent Pattern Hunter MCP server to explore how ideas emerge from many fragments and how those patterns can be transformed into actionable insights. This server provides a practical workflow for posing questions, letting fragments interact, detecting patterns, crystallizing insights, and weaving them into useful outputs.

How to use

To use the MCP server, you start the server locally and then connect with an MCP client or tooling that sends queries and receives structured reasoning results. You will submit a question or task, and the server will perform five phases: explode the idea into fragments, let fragments interact, detect emergent patterns, crystallize those patterns into insights, and weave the final answer. You can enable visualization to see intermediate steps and patterns. Use the client to request intermediate results for debugging or to study how the reasoning unfolds.

How to install

# Clone the repository
git clone https://github.com/yourusername/eph-mcp.git
cd eph-mcp

# Install dependencies
pip install -r requirements.txt
python -m spacy download en_core_web_sm

# Quick test
python quickstart.py

Start the MCP server with Python to expose the MCP endpoint on the default port. The server is ready to accept connections from MCP clients.

Configuration and usage notes

{
  "explosion": {
    "n_fragments": 100,
    "temperature": 1.5,
    "embedding_model": "all-MiniLM-L6-v2"
  },
  "interaction": {
    "iterations": 150,
    "initial_temperature": 1.0,
    "cooling_rate": 0.995
  },
  "detection": {
    "min_pattern_size": 3,
    "pattern_threshold": 0.5
  },
  "crystallization": {
    "confidence_threshold": 0.5,
    "novelty_threshold": 0.3
  },
  "weaving": {
    "max_insights": 5,
    "coherence_threshold": 0.6
  }
}

The server exposes tools for thinking emergently, pattern analysis, and reasoning history. You will typically run the server and then use a client tool to issue queries and retrieve results.

Testing and examples

# Basic tests
python tests/test_basic.py

# Full test suite (if available)
pytest tests/

python examples/usage_examples.py

Available tools

think_emergently

Main reasoning tool that applies the full EPH process to a given query, returning a structured reasoning trace and final answer.

analyze_patterns

Analyzes text to identify emergent patterns without performing full reasoning, useful for quick pattern checks.

compare_thoughts

Compares multiple ideas to identify relationships, contradictions, and harmonies across thoughts.

reasoning_history

Access past reasoning sessions to analyze previous results and patterns.