HUMMBL MCP Server

Instructions

Implementation Reference

• src/tools/methodology.ts:39-66 (handler)

  The asynchronous handler function for the 'get_methodology' MCP tool. It retrieves the Self-Dialectical methodology using getSelfDialecticalMethodology(), handles errors, and returns the data as both text (JSON) and structured content.

    async () => {
      const result = getSelfDialecticalMethodology();
  
      if (!isOk(result)) {
        return {
          content: [
            {
              type: "text" as const,
              text: `Unable to retrieve Self-Dialectical methodology: ${result.error.type}`,
            },
          ],
          isError: true as const,
        };
      }
  
      const methodology = result.value;
  
      return {
        content: [
          {
            type: "text" as const,
            text: JSON.stringify(methodology, null, 2),
          },
        ],
        structuredContent: methodology as unknown as { [key: string]: unknown },
      };
    }
  );

• src/tools/methodology.ts:15-38 (schema)

  Zod schemas defining the input (empty) and output structure for the 'get_methodology' tool, matching the DialecticalMethodology type.

  {
    title: "Get Self-Dialectical AI Methodology",
    description:
      "Retrieve the canonical Self-Dialectical AI Systems methodology with HUMMBL Base120 mappings.",
    inputSchema: z.object({}),
    outputSchema: z.object({
      id: z.string(),
      title: z.string(),
      version: z.string(),
      summary: z.string(),
      documentUrl: z.string().optional(),
      totalPages: z.number().optional(),
      modelsReferenced: z.array(z.string()),
      stages: z.array(
        z.object({
          stage: z.enum(["thesis", "antithesis", "synthesis", "convergence", "meta_reflection"]),
          title: z.string(),
          description: z.string(),
          modelCodes: z.array(z.string()),
        })
      ),
      metaModels: z.array(z.string()),
    }),
  },

• src/tools/methodology.ts:13-66 (registration)

  Direct registration of the 'get_methodology' tool on the MCP server within registerMethodologyTools, including name, schema, and handler.

  server.registerTool(
    "get_methodology",
    {
      title: "Get Self-Dialectical AI Methodology",
      description:
        "Retrieve the canonical Self-Dialectical AI Systems methodology with HUMMBL Base120 mappings.",
      inputSchema: z.object({}),
      outputSchema: z.object({
        id: z.string(),
        title: z.string(),
        version: z.string(),
        summary: z.string(),
        documentUrl: z.string().optional(),
        totalPages: z.number().optional(),
        modelsReferenced: z.array(z.string()),
        stages: z.array(
          z.object({
            stage: z.enum(["thesis", "antithesis", "synthesis", "convergence", "meta_reflection"]),
            title: z.string(),
            description: z.string(),
            modelCodes: z.array(z.string()),
          })
        ),
        metaModels: z.array(z.string()),
      }),
    },
    async () => {
      const result = getSelfDialecticalMethodology();
  
      if (!isOk(result)) {
        return {
          content: [
            {
              type: "text" as const,
              text: `Unable to retrieve Self-Dialectical methodology: ${result.error.type}`,
            },
          ],
          isError: true as const,
        };
      }
  
      const methodology = result.value;
  
      return {
        content: [
          {
            type: "text" as const,
            text: JSON.stringify(methodology, null, 2),
          },
        ],
        structuredContent: methodology as unknown as { [key: string]: unknown },
      };
    }
  );

• src/server.ts:23-23 (registration)

  Top-level call to registerMethodologyTools in the main server creation, which includes registration of 'get_methodology'.

  registerMethodologyTools(server);

• src/framework/self_dialectical.ts:82-84 (helper)

  Core helper function called by the tool handler to retrieve the static Self-Dialectical methodology data.

  export function getSelfDialecticalMethodology(): Result<DialecticalMethodology, DomainError> {
    return ok(SELF_DIALECTICAL_METHODOLOGY);
  }

• src/framework/self_dialectical.ts:16-80 (helper)

  Static constant defining the full methodology data returned by the tool, including stages and HUMMBL model references.

  export const SELF_DIALECTICAL_METHODOLOGY: DialecticalMethodology = {
    id: METHODOLOGY_ID,
    title: "Self-Dialectical AI Systems: A HUMMBL-Informed Methodology for Ethical Self-Correction",
    version: METHODOLOGY_VERSION,
    summary:
      "Enables AI systems to perform ethical self-correction through structured dialectical reasoning integrated with HUMMBL Base120 mental models.",
    documentUrl: undefined,
    totalPages: 60,
    modelsReferenced: [
      "P1",
      "P2",
      "DE3",
      "IN11",
      "IN2",
      "IN10",
      "CO4",
      "CO1",
      "CO20",
      "RE11",
      "RE15",
      "RE1",
      "RE7",
      "RE3",
      "RE16",
      "SY19",
      "SY1",
    ],
    stages: [
      {
        stage: "thesis",
        title: "Stage 1: Thesis Generation",
        description: "Generate and structure the initial ethical position.",
        modelCodes: ["P1", "P2", "DE3"],
      },
      {
        stage: "antithesis",
        title: "Stage 2: Antithesis Development",
        description:
          "Critique thesis via adversarial reasoning, premortem analysis, and red teaming.",
        modelCodes: ["IN11", "IN2", "IN10"],
      },
      {
        stage: "synthesis",
        title: "Stage 3: Synthesis Formation",
        description:
          "Integrate thesis and antithesis into a higher-order resolution across perspectives.",
        modelCodes: ["CO4", "CO1", "CO20"],
      },
      {
        stage: "convergence",
        title: "Stage 4: Convergence Testing",
        description:
          "Test and refine synthesis via calibration loops, convergence/divergence, and recursive improvement.",
        modelCodes: ["RE11", "RE15", "RE1"],
      },
      {
        stage: "meta_reflection",
        title: "Stage 5: Meta-Reflection",
        description:
          "Reflect on and improve the reasoning process itself via self-referential logic and meta-learning.",
        modelCodes: ["RE7", "RE3", "RE16"],
      },
    ],
    metaModels: ["SY19", "SY1"],
  };