I Built an AWS DevOps AI Agent Using Kiro Crew + MCP

Description and use

Kiro Crew combined with AWS DevOps Agent autonomously detects, investigates, and fixes silent failures in ECS, CodeBuild, and Lambda, cutting MTTR to 5‑7 minutes.

Problem

Silent failures in AWS services (ECS, CodeBuild, Lambda) that never trigger alarms, leading to prolonged MTTR and resource waste.

Outcome

Reduced MTTR from 1‑3 hours to 5‑7 minutes, eliminated silent failures, and automated incident response 24/7.

Workflow

• Configure Kiro Crew MCP server for AWS DevOps Agent via a single JSON block
• Set a cron job to run every 30 minutes
• Kiro Crew spawns 5 parallel subagents to check ECS, CI/CD, CloudWatch, DevOps Agent, and Lambda
• DevOps Agent performs root‑cause analysis and generates mitigation plan with CLI commands
• Kiro Crew executes the plan (or opens a PR for review)
• Verify service health and log the investigation

Evidence

Kiro Crew agent was awake. It spawned 5 parallel investigations, called AWS DevOps Agent for a health assessment, found the root cause across ECS, CodeBuild, CodePipeline, and Lambda, and flagged everything with severity‑prioritized fixes. By 3:24 AM, done.

Source

The architecture diagram shows Kiro Crew orchestrating subagents and DevOps Agent generating mitigation plans.

Source