AWS devopsagent documentation change
Summary
Added new 'Recommendation prioritization' section explaining how DevOps Agent ranks improvement suggestions
Security assessment
The new section describes operational prioritization of recommendations based on AI ranking and deterministic scoring. While incident prevention has indirect security benefits, there's no direct mention of security vulnerabilities, security features, or security-specific mechanisms. The content focuses on operational efficiency and workflow management.
Diff
diff --git a/devopsagent/latest/userguide/working-with-devops-agent-proactive-incident-prevention.md b/devopsagent/latest/userguide/working-with-devops-agent-proactive-incident-prevention.md index 9651e815f..6caf2540f 100644 --- a//devopsagent/latest/userguide/working-with-devops-agent-proactive-incident-prevention.md +++ b//devopsagent/latest/userguide/working-with-devops-agent-proactive-incident-prevention.md @@ -7 +7 @@ -How proactive incident prevention worksBenefitsAgent summaryControlling evaluationsManaging recommendationsAgent-ready specificationsImplementing recommendations +How proactive incident prevention worksBenefitsAgent summaryControlling evaluationsManaging recommendationsRecommendation prioritizationAgent-ready specificationsImplementing recommendations @@ -82,0 +83,45 @@ AWS DevOps Agent provides recommendations in the Improvements page where you can +## Recommendation prioritization + +AWS DevOps Agent automatically ranks your recommendations by priority to help you focus on the most impactful improvements first. The ranking considers your team’s specific context, operational patterns, and the severity of issues each recommendation addresses. + +### How prioritization works + +Each evaluation cycle, the agent ranks your active recommendations (those in a proposed or kept state) using a combination of: + + * **AI-powered ranking** – The agent assesses the relative importance of your top recommendations based on category relevance, incident severity, and operational impact. + + * **Deterministic scoring** – For larger backlogs, the agent applies a priority score based on incident frequency, severity patterns, and recency to ensure consistent ordering beyond the top-ranked items. + + + + +The ranked list appears in the Improvements page with a numeric rank position (1 being highest priority). Recommendations that have been discarded or implemented are not ranked. + +### Customizing priorities + +You can influence how the agent ranks recommendations by communicating your team’s priorities through the chat interface: + + * **Setting category preferences** – Tell the agent which recommendation categories matter most to your team (for example, "We prioritize observability improvements over infrastructure changes"). The agent stores these preferences and uses them in future ranking evaluations. + + * **Providing context** – Share information about upcoming projects, compliance requirements, or team focus areas. The agent incorporates this context when determining which recommendations should be prioritized. + + + + +To update your preferences, use the chat interface and describe your team’s priorities in natural language. The agent will confirm it has understood and will apply your preferences in the next evaluation cycle. + +### Rank stability + +Recommendation ranks may change between evaluation cycles when: + + * New recommendations are added that have higher priority than existing ones + + * Your team’s stated preferences change + + * New incident data strengthens or weakens the case for a recommendation + + + + +Recommendations that you have already marked as Keep retain their position in your backlog regardless of rank changes, ensuring your workflow is not disrupted. +