Mac Shell MCP Server

Instructions

Implementation Reference

• src/index.ts:362-388 (handler)

  Primary MCP tool handler: validates input with Zod schema, maps security level string to enum, delegates to CommandService.updateSecurityLevel, returns confirmation message.
  private async handleUpdateSecurityLevel(args: unknown) { const schema = z.object({ command: z.string(), securityLevel: z.enum(['safe', 'requires_approval', 'forbidden']), }); const { command, securityLevel } = schema.parse(args); // Map string security level to enum const securityLevelEnum = securityLevel === 'safe' ? CommandSecurityLevel.SAFE : securityLevel === 'requires_approval' ? CommandSecurityLevel.REQUIRES_APPROVAL : CommandSecurityLevel.FORBIDDEN; this.commandService.updateSecurityLevel(command, securityLevelEnum); return { content: [ { type: 'text', text: `Security level for command '${command}' updated to '${securityLevel}'`, }, ], }; }
• src/services/command-service.ts:226-232 (helper)

  Core logic implementation: retrieves the whitelist entry for the command, updates its securityLevel, and stores it back in the whitelist Map.
  public updateSecurityLevel(command: string, securityLevel: CommandSecurityLevel): void { const entry = this.whitelist.get(command); if (entry) { entry.securityLevel = securityLevel; this.whitelist.set(command, entry); } }
• src/index.ts:142-160 (registration)

  Tool registration in the ListTools response: defines name, description, and input schema for MCP protocol compliance.
  { name: 'update_security_level', description: 'Update the security level of a whitelisted command', inputSchema: { type: 'object', properties: { command: { type: 'string', description: 'The command to update', }, securityLevel: { type: 'string', enum: ['safe', 'requires_approval', 'forbidden'], description: 'New security level for the command', }, }, required: ['command', 'securityLevel'], }, },
• src/index.ts:363-366 (schema)

  Zod schema used in handler for runtime input validation, matching the declared inputSchema.
  const schema = z.object({ command: z.string(), securityLevel: z.enum(['safe', 'requires_approval', 'forbidden']), });
• src/services/command-service.ts:11-18 (schema)

  Type definition enum for CommandSecurityLevel used throughout the implementation.
  export enum CommandSecurityLevel { /** Safe commands that can be executed without approval */ SAFE = 'safe', /** Commands that require approval before execution */ REQUIRES_APPROVAL = 'requires_approval', /** Commands that are explicitly forbidden */ FORBIDDEN = 'forbidden', }