Cut DevOps incident triage from an hour to under 15 minutes: have an AI agent pull logs, service health and recent code changes, then post a structured root-cause summary to Slack before anyone opens a terminal: AI agent prompt
On Nov 2, 2026 AWS published a Circle K case study: the convenience retailer (17,200+ stores) built an Amazon Bedrock agentic triage pipeline for Grafana alerts. When an alert fires, specialized agents check ECS/CloudWatch health, OpenSearch logs/traces and GitHub code/commits in parallel, then an analysis step posts probable root cause, evidence, owner and next steps to Slack in 5–15 minutes. Mean triage time fell from 45–60 minutes to under 15 minutes (67–75% reduction). The team later moved the same logic to Bedrock AgentCore Runtime for longer investigations; humans still approve any remediation.
The prompt
1. List the tools your on-call already opens for every alert (logs, metrics, traces, deploy history, runbooks) and pick one alert type that always burns 30+ minutes of context gathering. 2. Give a coding/ops agent (Amazon Bedrock AgentCore, Claude, Cursor Automations or similar) read-only access to those sources plus a Slack channel; no restart or deploy permissions yet. 3. Instruct it: "When this alert fires, gather recent logs, service health, related traces and the last relevant commits. Return: probable root cause, evidence with timestamps, affected service/owner, and recommended next steps. Say what you could not verify." 4. Have it post that structured summary to Slack within a fixed time budget (for example 15 minutes) instead of raw alert noise. 5. Compare 10 incidents: time-to-first-useful-context with the agent vs without. Only after that accuracy check, let a human approve simple remediations (restart, scale, rollback) from Slack buttons.
How to use it
- Copy the prompt above.
- Replace every [bracket] with your own details (or fill in the blanks on AgentPlaybook Real Estate, which remembers your answers for the next prompt).
- Paste it into Amazon Bedrock / AgentCore (or Claude/Cursor agent), Grafana or your monitor, CloudWatch/OpenSearch or equivalent logs and review the answer before you use it.
Works with: Amazon Bedrock / AgentCore (or Claude/Cursor agent), Grafana or your monitor, CloudWatch/OpenSearch or equivalent logs, GitHub, Slack
Real-world example: aws.amazon.com/blogs/machine-learning/how-circle-k-accelerated-devops-
Featured in: AI prompts to summarize meetings, notes, documents and reports
Before you use it: what to check
Already built into this prompt
- Only after that accuracy check, let a human approve simple remediations (restart, scale, rollback) from Slack buttons.
- Give a coding/ops agent (Amazon Bedrock AgentCore, Claude, Cursor Automations or similar) read-only access to those sources plus a Slack channel; no restart or deploy permissions yet.
- Say what you could not verify." 4.
Check yourself
- Check every figure, price and total against your own records before anything is sent, posted or booked. A person approves refunds, discounts, payments and price changes.
- Read and edit the draft before it reaches a customer or goes public. If you later automate this prompt, start with "draft and notify me", not "send automatically".
- Check names, dates, numbers and any claim against your own source material before you use the answer. AI can misread or invent details.
Using an AI agent? Download this prompt as a SKILL.md agent skill with these checks built in, or get the .zip for Claude.ai. How to install skills.
Want to make this prompt even better? Grade it with the free prompt grader or build your own with the AI prompt generator.
32 copy-paste agent prompts (PDF) plus 5 new prompts every week.
Get it freeMore IT & Internal Help prompts
See all IT & Internal Help prompts →