/investigate

Use the investigate skill workflow.

1. If no AgentSpace is yet known in this session, call aws___call_aws(cli_command="aws devops-agent list-agent-spaces --region us-east-1") and pick the right one (consult .claude/aws-devops-agent.md if present; ask the user if multiple plausible candidates).
2. Gather local context — git log --oneline -10, dependency manifest, relevant IaC, the error/log the user is looking at — and pack it into the --description parameter.
3. aws___call_aws(cli_command="aws devops-agent create-backlog-task --agent-space-id SPACE_ID --task-type INVESTIGATION --title '$ARGUMENTS' --priority HIGH --description '<local context + question>' --region us-east-1").
4. Tell the user investigations take 5–8 minutes and that you'll keep them posted.
5. Poll aws___call_aws(cli_command="aws devops-agent get-backlog-task --agent-space-id SPACE_ID --task-id TASK_ID --region us-east-1") every 30–45s; when IN_PROGRESS and executionId is available, paginate aws___call_aws(cli_command="aws devops-agent list-journal-records --agent-space-id SPACE_ID --execution-id EXEC_ID --order ASC --next-token TOKEN --region us-east-1"); summarize each new record to the user using the emoji prefixes from the investigate skill.
6. On COMPLETED: pull the consolidated summary (--order DESC --max-results 10), then aws___call_aws(cli_command="aws devops-agent list-recommendations --agent-space-id SPACE_ID --task-id TASK_ID --region us-east-1") → aws___call_aws(cli_command="aws devops-agent get-recommendation --agent-space-id SPACE_ID --recommendation-id REC_ID --region us-east-1") for each. Show the user the proposed fix; do not auto-apply IaC or commands.

If $ARGUMENTS is empty, ask the user for a one-line incident description first.