Aurora/RDS Incident Business Context (Skill)¶
Enriches an Amazon Aurora or Amazon RDS incident investigation with business context — dependent services, estimated cost of downtime, compliance reporting exposure, and recent-change correlation — so AWS DevOps Agent's technical root-cause findings come with an answer to "who does this affect and what does it cost us."
⚠️ This skill is sample code, not intended for production use without additional review and testing. Users should validate in a non-production environment first.
What it does¶
DevOps Agent's built-in investigation tools (CloudWatch, Performance Insights, RDS/Aurora logs) answer what is technically wrong. This skill adds the business half of the incident report: which teams/services are affected, a defensible dollar estimate of downtime cost, which compliance reporting obligations could apply (for the compliance owner to assess), and whether a recent change correlates with the incident.
It is evidence-driven, not generative — it tells the agent where to look (resource tags, AWS Config history, CloudTrail, customer-supplied dependency maps) and how to reason about what it finds. It does not invent dependency graphs, dollar figures, or compliance frameworks. If the evidence isn't tagged or configured in the account, the skill says so and asks the operator, rather than guessing.
Prerequisites¶
For the skill to find real evidence (rather than reporting gaps), the target account should have some of the following in place:
- IAM permissions on the DevOps Agent role. All four are read-only.
AIDevOpsAgentAccessPolicyalready coversrds:ListTagsForResource,tag:GetResourcesandcloudtrail:LookupEvents. It doesn't coverconfig:GetResourceConfigHistory, which the recent-change check uses. Add it with theEnableAuroraIncidentBusinessContextparameter ofdevops-agent-skill-policies.yaml. Without it, the skill falls back to CloudTrail and says that Config history wasn't readable. - Resource tags on the Aurora/RDS resource, e.g.
DependentServices,OnCallTeam,RevenuePerMinute,SLAThresholdMinutes,DataClassification,Compliance. None are required for the skill to activate — it degrades gracefully to "no evidence found" per category.
How to use it with DevOps Agent¶
Install the skill in your Agent Space (Investigations and/or Chat subagents, depending on where you want business-context enrichment available). It activates automatically once DevOps Agent is investigating an Aurora/RDS alarm and the reasoning trace calls for business impact.
You can also invoke it on demand in chat, e.g.:
Investigate the Aurora cluster
example-cluster— theexample-cluster-connections-highalarm is firing. Once you've found the technical root cause, tell me who's affected, what this costs us, and whether we have a compliance reporting deadline.
Origin¶
Adapted from the business-context enrichment logic prototyped in the Aurora MySQL Incident Investigation with AWS DevOps Agent demo (GenAI Ops Demo Library), currently proposed in aws-samples/sample-aws-genai-ops-demos#95. That demo originally shipped this as a Lambda-backed mock MCP server returning synthetic example data (fictional dependency graph, a fixed revenue figure, canned compliance thresholds) for demonstration purposes. This skill generalizes that same reasoning into an evidence-based framework suitable for real accounts, with no infrastructure to deploy.
The demo above shows this skill's reasoning in action against a live incident (connection storm, CPU spike, deadlock, or failover on a real Aurora MySQL cluster) — see its README for a runnable, end-to-end walkthrough.