MCP Rand

Instructions

Implementation Reference

• src/handlers/generate-gaussian.handler.ts:56-69 (handler)

  The main handler function for the 'generate_gaussian' tool. It calls the internal generateGaussian() helper and returns the result as MCP tool content (text).

  export const generateGaussianHandler = async (
    _request: CallToolRequest
  ): Promise<CallToolResult> => {
    const gaussian = generateGaussian();
    
    return {
      content: [
        {
          type: 'text',
          text: gaussian.toString()
        }
      ]
    };
  };

• src/handlers/generate-gaussian.handler.ts:6-13 (schema)

  Tool schema/specification defining name, description, and empty input schema for no parameters.

  export const toolSpec = {
    name: 'generate_gaussian',
    description: 'Generate a random number following a Gaussian (normal) distribution between 0 and 1',
    inputSchema: {
      type: 'object' as const,
      properties: {}
    }
  };

• src/index.ts:21-21 (registration)

  Registers the generateGaussianHandler under the 'tools/call' method for the tool name 'generate_gaussian' in the MCP server.

  registry.register('tools/call', 'generate_gaussian', generateGaussianHandler as Handler);

• src/handlers/generate-gaussian.handler.ts:16-35 (helper)

  Core helper implementing Box-Muller transform with error function to generate Gaussian random number normalized to [0,1].

  function generateGaussian(): number {
    let u1 = 0;
    let u2 = 0;
    
    // Avoid u1 being zero
    do {
      u1 = Math.random();
      u2 = Math.random();
    } while (u1 <= Number.EPSILON);
  
    const z0 = Math.sqrt(-2.0 * Math.log(u1)) * Math.cos(2.0 * Math.PI * u2);
    
    // Convert from standard normal distribution (mean 0, std dev 1)
    // to a value between 0 and 1 using the error function (erf)
    // We add 1 and divide by 2 to shift from [-1,1] to [0,1]
    const normalized = (erf(z0 / Math.SQRT2) + 1) / 2;
    
    // Clamp to [0,1] in case of floating point errors
    return Math.max(0, Math.min(1, normalized));
  }

• src/handlers/generate-gaussian.handler.ts:38-54 (helper)

  Approximation of the error function (erf) used in generateGaussian for normalizing the distribution.

  // Abramowitz and Stegun approximation (maximum error: 1.5×10−7)
  function erf(x: number): number {
    const sign = Math.sign(x);
    x = Math.abs(x);
  
    const p = 0.3275911;
    const a1 = 0.254829592;
    const a2 = -0.284496736;
    const a3 = 1.421413741;
    const a4 = -1.453152027;
    const a5 = 1.061405429;
  
    const t = 1.0 / (1.0 + p * x);
    const poly = t * (a1 + t * (a2 + t * (a3 + t * (a4 + t * a5))));
  
    return sign * (1 - poly * Math.exp(-x * x));
  }