exec-aide/lib/constructs/socket-mode.ts

import { Construct } from 'constructs';
import * as cdk from 'aws-cdk-lib';
import * as ec2 from 'aws-cdk-lib/aws-ec2';
import * as ecs from 'aws-cdk-lib/aws-ecs';
import * as ecr from 'aws-cdk-lib/aws-ecr';
import * as ecrAssets from 'aws-cdk-lib/aws-ecr-assets';
import * as logs from 'aws-cdk-lib/aws-logs';
import * as iam from 'aws-cdk-lib/aws-iam';
import * as dynamodb from 'aws-cdk-lib/aws-dynamodb';
import * as cloudwatch from 'aws-cdk-lib/aws-cloudwatch';
import * as cloudwatchActions from 'aws-cdk-lib/aws-cloudwatch-actions';
import * as sns from 'aws-cdk-lib/aws-sns';
import * as path from 'path';

export interface SocketModeProps {
  table: dynamodb.ITable;
  conversationFnArn: string;
  /**
   * Shared site-alerts SNS topic for CloudWatch ALARM actions. Imported once at
   * the stack level (sns.Topic.fromTopicArn) and injected here, mirroring how
   * the table is wired into this construct.
   */
  alarmTopic: sns.ITopic;
}

export class SocketModeConstruct extends Construct {
  constructor(scope: Construct, id: string, props: SocketModeProps) {
    super(scope, id);

    const account = cdk.Stack.of(this).account;
    const region = cdk.Stack.of(this).region;

    // ── VPC ────────────────────────────────────────────────────

    const vpc = new ec2.Vpc(this, 'Vpc', {
      vpcName: 'exec-aide',
      ipAddresses: ec2.IpAddresses.cidr('10.30.0.0/16'),
      maxAzs: 1,
      natGateways: 0,
      subnetConfiguration: [
        {
          cidrMask: 24,
          name: 'exec-aide-listener',
          subnetType: ec2.SubnetType.PUBLIC,
        },
      ],
    });

    const sg = new ec2.SecurityGroup(this, 'ListenerSG', {
      vpc,
      description: 'exec-aide listener - outbound only',
      allowAllOutbound: true,
    });

    // ── ECR ────────────────────────────────────────────────────

    new ecr.Repository(this, 'ListenerRepo', {
      repositoryName: 'exec-aide-listener',
      imageScanOnPush: true,
      lifecycleRules: [
        {
          maxImageCount: 5,
          description: 'Keep last 5 images',
        },
      ],
      removalPolicy: cdk.RemovalPolicy.RETAIN,
    });

    // ── ECS Cluster + Task Definition ──────────────────────────

    const cluster = new ecs.Cluster(this, 'Cluster', {
      clusterName: 'exec-aide',
      vpc,
    });

    const taskDef = new ecs.FargateTaskDefinition(this, 'TaskDef', {
      family: 'exec-aide-listener',
      cpu: 256,
      memoryLimitMiB: 512,
      runtimePlatform: {
        cpuArchitecture: ecs.CpuArchitecture.ARM64,
        operatingSystemFamily: ecs.OperatingSystemFamily.LINUX,
      },
    });

    taskDef.addContainer('listener', {
      image: ecs.ContainerImage.fromAsset(
        path.join(__dirname, '../../listener'),
        // Explicit platform: the task is ARM64, and without this an amd64 CI
        // runner builds an amd64 image even with QEMU enabled — tasks then
        // crash-loop with "exec format error" (50 failed starts on 2026-06-03).
        { platform: ecrAssets.Platform.LINUX_ARM64 },
      ),
      essential: true,
      environment: {
        TABLE_NAME: props.table.tableName,
        SECRET_SLACK: 'exec-aide/slack-credentials',
        SSM_PREFIX: '/exec-aide',
        CONVERSATION_FN_ARN: props.conversationFnArn,
      },
      logging: ecs.LogDrivers.awsLogs({
        streamPrefix: 'listener',
        logGroup: new logs.LogGroup(this, 'ListenerLogGroup', {
          logGroupName: '/ecs/exec-aide-listener',
          retention: logs.RetentionDays.TWO_MONTHS,
        }),
      }),
    });

    // ── Task role IAM ──────────────────────────────────────────

    props.table.grantReadWriteData(taskDef.taskRole);

    taskDef.taskRole.addToPrincipalPolicy(new iam.PolicyStatement({
      actions: ['secretsmanager:GetSecretValue'],
      resources: [
        `arn:aws:secretsmanager:${region}:${account}:secret:exec-aide/slack-credentials-*`,
      ],
    }));

    taskDef.taskRole.addToPrincipalPolicy(new iam.PolicyStatement({
      actions: ['ssm:GetParametersByPath', 'ssm:GetParameter'],
      resources: [
        `arn:aws:ssm:${region}:${account}:parameter/exec-aide`,
        `arn:aws:ssm:${region}:${account}:parameter/exec-aide/*`,
      ],
    }));

    taskDef.taskRole.addToPrincipalPolicy(new iam.PolicyStatement({
      actions: ['lambda:InvokeFunction'],
      resources: [props.conversationFnArn],
    }));

    // ── Fargate Service ────────────────────────────────────────
    // Assigned to a const (no logical-id change vs. the prior anonymous
    // construct — id 'Service' is unchanged) so the service-level CloudWatch
    // metrics below can reference it.
    const service = new ecs.FargateService(this, 'Service', {
      serviceName: 'exec-aide-listener',
      cluster,
      taskDefinition: taskDef,
      desiredCount: 1,
      assignPublicIp: true,
      securityGroups: [sg],
      vpcSubnets: { subnetType: ec2.SubnetType.PUBLIC },
    });

    // ── CloudWatch Alarms ──────────────────────────────────────
    // ALARM-only (no OK / InsufficientData action); treatMissingData
    // NOT_BREACHING. Shared site-alerts action injected via props.
    // CPU and Memory come from AWS/ECS service metrics (no Container Insights
    // required). The RunningTaskCount alarm — which DOES require Container
    // Insights — is added in a separate, sign-off-gated commit.
    const alarmAction = new cloudwatchActions.SnsAction(props.alarmTopic);

    new cloudwatch.Alarm(this, 'ListenerCpuHigh', {
      alarmName: 'exec-aide-listener-cpu-high',
      alarmDescription: 'exec-aide-listener Fargate service CPU utilization is high',
      metric: service.metricCpuUtilization({
        period: cdk.Duration.minutes(5),
        statistic: 'Average',
      }),
      threshold: 80,
      comparisonOperator: cloudwatch.ComparisonOperator.GREATER_THAN_THRESHOLD,
      evaluationPeriods: 3,
      datapointsToAlarm: 2,
      treatMissingData: cloudwatch.TreatMissingData.NOT_BREACHING,
    }).addAlarmAction(alarmAction);

    new cloudwatch.Alarm(this, 'ListenerMemoryHigh', {
      alarmName: 'exec-aide-listener-memory-high',
      alarmDescription: 'exec-aide-listener Fargate service memory utilization is high',
      metric: service.metricMemoryUtilization({
        period: cdk.Duration.minutes(5),
        statistic: 'Average',
      }),
      threshold: 80,
      comparisonOperator: cloudwatch.ComparisonOperator.GREATER_THAN_THRESHOLD,
      evaluationPeriods: 3,
      datapointsToAlarm: 2,
      treatMissingData: cloudwatch.TreatMissingData.NOT_BREACHING,
    }).addAlarmAction(alarmAction);
  }
}