Azure Omni-Tool MCP Server

Instructions

Implementation Reference

• src/tools/context-retriever.ts:116-141 (handler)

  Main handler function for get_azure_context tool. Dispatches to specific query handlers (subscriptions, resource_groups, resources, custom) based on input.query_type, handles caching, logging, auditing, and returns ContextResponse.

  export async function handleGetAzureContext(input: GetAzureContextInput): Promise<ContextResponse> {
      const { query_type } = input;
      const operator = getOperatorInfo();
      const audit = createAuditContext(`context:${query_type}`, 'low', 'query');
  
      logger.debug('Context query', { query_type, subscription: input.subscription_id, rg: input.resource_group });
  
      const handler = queryHandlers.get(query_type);
      if (!handler) {
          return { query_type, success: false, error: `Unknown query type: ${query_type}` };
      }
  
      const response = await handler(input);
      response.correlation_id = audit.correlationId;
      response.operator = operator;
  
      if (response.success) {
          await audit.logSuccess();
          logger.info('Context query succeeded', { query_type, count: response.count, cached: response.cached });
      } else {
          await audit.logFailure(response.error || 'Unknown');
          logger.warn('Context query failed', { query_type, error: response.error });
      }
  
      return response;
  }

• src/tools/context-retriever.ts:10-16 (schema)

  Zod schema defining input validation for get_azure_context tool, including query_type (required), optional filters, custom_query for KQL, and cache bypass.

  export const GetAzureContextSchema = z.object({
      query_type: z.enum(['subscriptions', 'resource_groups', 'resources', 'custom']).describe('Type of context query'),
      subscription_id: z.string().optional().describe('Subscription to scope query'),
      resource_group: z.string().optional().describe('Resource group filter'),
      custom_query: z.string().optional().describe('KQL query for Resource Graph'),
      bypass_cache: z.boolean().optional().default(false).describe('Force fresh data'),
  });

• src/index.ts:28-32 (registration)

  Tool registration in the toolRegistry Map, linking name 'get_azure_context' to its tool metadata, schema, and handler wrapper.

  ['get_azure_context', {
      tool: getAzureContextTool,
      schema: GetAzureContextSchema,
      handler: args => handleGetAzureContext(GetAzureContextSchema.parse(args))
  }],

• src/tools/context-retriever.ts:143-165 (helper)

  Tool metadata object with name, detailed description, and JSON schema for MCP tool listing.

  export const getAzureContextTool = {
      name: 'get_azure_context',
      description: `Retrieves Azure environment context.
  
  TYPES: subscriptions, resource_groups, resources, custom (KQL)
  
  CACHING: 5min default, bypass_cache=true for fresh data
  
  EXAMPLES: 
  - Find VMs: custom_query = "Resources | where type == 'microsoft.compute/virtualmachines'"
  - Find by tag: custom_query = "Resources | where tags.env == 'prod'"`,
      inputSchema: {
          type: 'object',
          properties: {
              query_type: { type: 'string', enum: ['subscriptions', 'resource_groups', 'resources', 'custom'] },
              subscription_id: { type: 'string', description: 'Optional subscription scope' },
              resource_group: { type: 'string', description: 'Optional resource group filter' },
              custom_query: { type: 'string', description: 'KQL query for Resource Graph' },
              bypass_cache: { type: 'boolean', description: 'Force fresh data', default: false },
          },
          required: ['query_type'],
      },
  };