Event Horizon

An MCP server implementation that provides evolutionary solution generation and optimization capabilities for LLMs. Event Horizon enables LLMs to evolve solutions across multiple generations using consistency check evaluations and genetic algorithm principles.

javascript

GitHub Stars

javascript

Language

7 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "manasp21-eventhorizon": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "event-horizon-mcp"
      ],
      "env": {
        "DISABLE_EVOLUTION_LOGGING": "true"
      }
    }
  }
}

You can run and interact with the Event Horizon MCP Server to evolve and optimize solutions for complex problems using multi-generational genetic-like processes. You configure problem statements, run generations of candidate solutions, score them against criteria, and receive guided crossover insights to improve future generations.

How to use

You operate the Event Horizon MCP Server through an MCP client. Start a new evolution by defining your problem statement and a set of consistency checks. Add initial solutions for generation 0, then score each solution against every check. Use the crossover guidance to identify how to combine best aspects from different solutions, and create the next generation. Monitor progress and repeat until convergence or a maximum number of generations is reached. Typical actions you perform include starting evolution, adding solutions, scoring, evolving generations, and querying the current status.

How to install

Prerequisites: you must have Node.js and npm installed on your machine. You may also choose to run the server via Docker or directly as a local Node process.

# Prerequisites
node -v
npm -v

# Install the server locally (dev/build steps are handled in the next commands)
# If you want to build from source first
git clone https://github.com/manasp21/EventHorizon.git
cd EventHorizon
npm install
npm run build

# Run the server directly with Node (adjust the path to the built entry as needed)
node dist/index.js
```} ,{

Or run the server using Docker with a prebuilt image. Build and run the container as shown below.

# Build the Docker image
docker build -t event-horizon-mcp .

# Run the container
docker run --rm -i event-horizon-mcp
```} ,{

Configuration and notes

Configure the MCP server as part of your client setup. You can reference the provided sample configuration to run the server locally or via Docker. Use environment variables to customize behavior, such as disabling progress logging.

Environment variables

DISABLE_EVOLUTION_LOGGING: Set to "true" to disable evolution progress logging.

Tools and endpoints

The server exposes core tools that manage the evolution workflow, including starting evolution, adding solutions, scoring, evolving generations, and querying status.

Advanced configuration

Configuration options control how evolution runs. You can adjust default values for population size, maximum generations, and the convergence threshold. You can also choose to disable logging.

Docker and client integrations are supported to run and connect the MCP server from your preferred environment.

Usage example with the evolution life cycle

Start Evolution: Define your problem and checks, then initialize the evolution workflow.
Add Initial Solutions: Provide several candidate solutions addressing the problem.
Score Solutions: Evaluate each candidate against the consistency checks and capture the scores and reasoning.
Evolve Generation: Obtain crossover recommendations and create the next generation based on the guidance.
Repeat: Continue scoring and evolving until you reach convergence or the maximum number of generations.

Available tools

start_evolution

Initialize an evolutionary solution system with a problem and consistency checks.

add_solution

Add a new solution to the current generation, optionally referencing parent solutions for crossovers.

score_solution

Score a solution against a specific consistency check and optionally provide reasoning for the score.

evolve_generation

Evolve to the next generation using crossover analysis and return recommendations for creating new solutions.

get_evolution_status

Retrieve the current progress, statistics, and best solution found so far.

Built by

VeilStrat

AI signals for GTM teams