landing_zone_design

Instructions

Implementation Reference

• src/tools/architecture/landing-zone-design.ts:55-68 (handler)

  The handler function that executes the landing_zone_design tool logic. It uses runTool() to validate args via Zod schema, builds a prompt via landingZoneTemplate(), calls the Anthropic API with the ARCHITECTURE_SYSTEM system prompt and token budget, then returns the text response.

  export async function handleLandingZone(args: unknown): Promise<string> {
    return runTool('landing_zone_design', args, Schema, async (params) => {
      const prompt = landingZoneTemplate({ ...params, cssp: params.cssp ?? 'azure-gcc-high' });
  
      const response = await anthropic.messages.create({
        model: MODEL,
        max_tokens: getTokenBudget('landing_zone_design'),
        system: ARCHITECTURE_SYSTEM,
        messages: [{ role: 'user', content: prompt }],
      });
  
      return response.content[0].type === 'text' ? response.content[0].text : '';
    });
  }

• src/tools/architecture/landing-zone-design.ts:44-53 (schema)

  Zod validation schema for landing_zone_design inputs. Validates missionType, dataClassification, userBase, targetImpactLevel (required), plus optional estimatedUsers, connectedToNIPR, existingEnclaves, and cssp (default azure-gcc-high).

  const Schema = z.object({
    missionType: z.enum(['combat-support', 'admin-backoffice', 'legal-services', 'healthcare', 'intelligence-analytics', 'logistics', 'communications']),
    dataClassification: z.enum(['unclassified', 'cui', 'fouo', 'secret']),
    userBase: z.enum(['conus', 'oconus', 'both', 'disconnected']),
    targetImpactLevel: z.enum(['fedramp-moderate', 'fedramp-high', 'il4', 'il5']),
    estimatedUsers: z.number().optional(),
    connectedToNIPR: z.boolean().optional(),
    existingEnclaves: z.string().max(500).optional(),
    cssp: z.enum(['azure-government', 'azure-gcc-high']).default('azure-gcc-high'),
  });

• src/tools/index.ts:39-39 (registration)

  Tool registration: landingZoneTool is included in the allTools array at index position for architecture tools.

  landingZoneTool,

• src/tools/index.ts:69-69 (registration)

  Tool dispatch: the handleToolCall switch routes 'landing_zone_design' to handleLandingZone(args).

  case 'landing_zone_design':   return handleLandingZone(args);

• src/prompts/templates.ts:93-127 (helper)

  The prompt template function landingZoneTemplate() that constructs the detailed architecture prompt passed to the Anthropic API, requesting hub-spoke topology, subscription layout, network CIDR, security services, identity architecture, Bicep scaffold, cost estimate, and ATO path.

  export function landingZoneTemplate(params: {
    missionType: string;
    dataClassification: string;
    userBase: string;
    targetImpactLevel: string;
    estimatedUsers?: number;
    connectedToNIPR?: boolean;
    existingEnclaves?: string;
    cssp: string;
  }): string {
    return `Design a complete Azure Landing Zone for a government workload with these parameters:
  
  Mission Type: ${params.missionType}
  Data Classification: ${params.dataClassification}
  User Base: ${params.userBase}
  Target Impact Level: ${params.targetImpactLevel}
  CSP Level: ${params.cssp}
  ${params.estimatedUsers ? `Estimated Users: ${params.estimatedUsers}` : ''}
  ${params.connectedToNIPR !== undefined ? `Connected to NIPR: ${params.connectedToNIPR}` : ''}
  ${params.existingEnclaves ? `Existing Enclaves: ${params.existingEnclaves}` : ''}
  
  Provide:
  1. Management Group hierarchy (with naming convention)
  2. Subscription topology
  3. Hub-spoke network design (CIDR ranges, subnet layout)
  4. Required Azure security services
  5. Identity architecture (Entra ID + MFA + PIM)
  6. Connectivity model
  7. Bicep scaffold structure
  8. Estimated monthly cost range
  9. ATO path recommendation
  10. What is typically missed on ${params.missionType} deployments
  
  Be specific to the mission type — ${params.missionType} has unique requirements.`;
  }