Skip to content

Operation Phase -- Stage Reference (4.1-4.7)

Phase Overview

The Operation phase is the fifth of five phases in the AI-DLC lifecycle. It takes the built, tested software from Construction and moves it through deployment, monitoring, incident preparedness, performance validation, and continuous optimization. It covers seven stages (4.1 through 4.7) that span pipeline configuration, environment provisioning, deployment execution, observability, incident response, performance validation, and feedback collection.

All seven Operation stages are CONDITIONAL -- they execute based on the scope and execution plan. For example, mvp, poc, bugfix, and refactor scopes skip Operation entirely. The infra and security-patch scopes run a subset (deployment and environment stages).

All stages run inline (no subagents in the Operation phase). All stages follow stage-protocol.md for approval gates, question format, completion messages, and state tracking.


Stage Summary Table

Stage Name Execution Condition Lead Agent Support Agents Mode
4.1 Deployment Pipeline CONDITIONAL Execute when CD pipeline needs creation or significant modification aidlc-pipeline-deploy-agent (none) inline
4.2 Environment Provisioning CONDITIONAL Execute when AWS environments need provisioning or validation aidlc-aws-platform-agent aidlc-devsecops-agent, aidlc-compliance-agent inline
4.3 Deployment Execution CONDITIONAL Execute after deployment pipeline and environment are ready aidlc-pipeline-deploy-agent aidlc-developer-agent inline
4.4 Observability Setup CONDITIONAL Execute when monitoring, dashboards, alarms, or tracing need config aidlc-operations-agent (none) inline
4.5 Incident Response CONDITIONAL Execute when operational runbooks and incident response procedures needed aidlc-operations-agent (none) inline
4.6 Performance Validation CONDITIONAL Execute when NFR performance targets need validation under load aidlc-quality-agent (none) inline
4.7 Feedback & Optimization CONDITIONAL Execute when ongoing operational monitoring and optimization needed aidlc-operations-agent aidlc-aws-platform-agent inline

Multi-Agent Stages

Three Operation stages involve multiple agents:

  • 4.2 Environment Provisioning: aidlc-aws-platform-agent (lead) + aidlc-devsecops-agent (security posture validation) + aidlc-compliance-agent (data residency, regulatory controls)
  • 4.3 Deployment Execution: aidlc-pipeline-deploy-agent (lead) + aidlc-developer-agent (database migrations)
  • 4.7 Feedback & Optimization: aidlc-operations-agent (lead) + aidlc-aws-platform-agent (cost optimization, drift detection)

In all cases, the conductor invokes the lead agent first, then invokes support agents with the lead's output as context. The conductor performs every delegation; agents never invoke each other.


Stage 4.1: Deployment Pipeline Configuration

Metadata

Property Value
Stage 4.1
Phase Operation
Execution CONDITIONAL (skip if deployment pipeline already exists and is adequate)
Lead Agent aidlc-pipeline-deploy-agent
support_agents (none)
Inputs CI pipeline config from Stage 3.7, infrastructure design from Stage 3.4

Purpose

Configure the CD pipeline, deployment strategy, rollback procedures, and environment promotion gates.

Outputs

Artifact Description
cd-config.md CD pipeline configuration
deployment-strategy.md Deployment strategy (blue/green, canary, rolling), promotion gates
rollback-runbook.md Rollback procedures and runbook
deployment-pipeline-questions.md Clarifying questions with answers

Approval Gate

Strictly 2-option: Approve / Request Changes.


Stage 4.2: Environment Provisioning

Metadata

Property Value
Stage 4.2
Phase Operation
Execution CONDITIONAL (skip if environments already provisioned)
Lead Agent aidlc-aws-platform-agent
support_agents aidlc-devsecops-agent (security posture validation), aidlc-compliance-agent (data residency, regulatory controls)
Inputs Infrastructure design from Stage 3.4, CD pipeline config from Stage 4.1

Purpose

Provision and validate target AWS environments using Infrastructure as Code from Construction. The aidlc-devsecops-agent validates security posture and the aidlc-compliance-agent checks data residency and regulatory controls.

Outputs

Artifact Description
environment-inventory.md Provisioned environment inventory
validation-report.md Infrastructure validation report, health checks
environment-provisioning-questions.md Clarifying questions with answers

Approval Gate

Strictly 2-option: Approve / Request Changes.


Stage 4.3: Deployment Execution

Metadata

Property Value
Stage 4.3
Phase Operation
Execution CONDITIONAL (execute after deployment pipeline and environment are ready; skip if already deployed)
Lead Agent aidlc-pipeline-deploy-agent
support_agents aidlc-developer-agent (database migrations)
Inputs CD pipeline config from Stage 4.1, provisioned environments from Stage 4.2

Purpose

Execute the actual deployment: push artifacts through the pipeline, run smoke tests, validate health checks, and execute database migrations.

Outputs

Artifact Description
deployment-log.md Deployment execution log
smoke-test-results.md Smoke test results after deployment
health-check-report.md Health check validation report
deployment-execution-questions.md Pre-deployment check questions with answers

Approval Gate

Strictly 2-option: Approve / Request Changes.


Stage 4.4: Observability Setup

Metadata

Property Value
Stage 4.4
Phase Operation
Execution CONDITIONAL (skip if observability already configured)
Lead Agent aidlc-operations-agent
Inputs NFR design from Stage 3.3, infrastructure design from Stage 3.4, deployed application

Purpose

Configure monitoring, dashboards, alarms, SLO/SLI tracking, log queries, distributed tracing, and anomaly detection.

Outputs

Artifact Description
dashboards.md CloudWatch dashboard configurations
alarms.md Alarm definitions with severity, SNS routing, escalation
slo-config.md SLO/SLI tracking configuration
log-queries.md CloudWatch Logs Insights saved queries
tracing-config.md X-Ray tracing configuration
anomaly-config.md Anomaly detection configuration
observability-setup-questions.md Clarifying questions with answers

Notes

  • Produces the most artifacts of any Operation stage (6 content files + questions).
  • AWS-specific (CloudWatch, X-Ray, SNS) but patterns are transferable.

Stage 4.5: Incident Response & Runbook Generation

Metadata

Property Value
Stage 4.5
Phase Operation
Execution CONDITIONAL (skip for POCs or non-production deployments)
Lead Agent aidlc-operations-agent
Inputs Observability setup from Stage 4.4, NFR design from Stage 3.3, infrastructure design from Stage 3.4

Purpose

Generate operational runbooks, incident response plans, and escalation procedures.

Outputs

Artifact Description
runbooks.md SSM Automation runbook library
incident-plan.md Incident response plan (AWS Incident Manager integration)
escalation-matrix.md Escalation paths, on-call rotations, communication procedures
incident-response-questions.md Clarifying questions with answers

Stage 4.6: Performance Validation & Load Testing

Metadata

Property Value
Stage 4.6
Phase Operation
Execution CONDITIONAL (skip for POCs or non-performance-critical applications)
Lead Agent aidlc-quality-agent
Inputs NFR requirements from Stage 3.2, NFR design from Stage 3.3, observability data from Stage 4.4

Purpose

Design and execute load tests to validate NFR performance targets against the deployed application.

Outputs

Artifact Description
load-test-plan.md Load test plan with scenarios, tools, and configuration
test-results.md Performance test results (latency, throughput, error rates)
nfr-validation-matrix.md NFR target vs. actual validation matrix
performance-validation-questions.md Clarifying questions with answers

Stage 4.7: Continuous Feedback & Optimization

Metadata

Property Value
Stage 4.7
Phase Operation
Execution CONDITIONAL (skip for one-off deployments)
Lead Agent aidlc-operations-agent
support_agents aidlc-aws-platform-agent (cost optimization, drift detection)
Inputs All Operation phase artifacts, production monitoring data

Purpose

SLO compliance review, cost optimization analysis, infrastructure drift detection, and operational insights collection. This is the final stage of the entire AI-DLC workflow.

Outputs

Artifact Description
slo-report.md SLO compliance report, error budget burn rate
cost-analysis.md AWS Cost Explorer analysis, optimization recommendations
drift-report.md AWS Config drift detection report, Trusted Advisor review
feedback-loop.md Operational insights, improvement proposals, inputs to next Ideation cycle
feedback-optimization-questions.md Clarifying questions with answers

Approval Gate -- Three-Option (Unique)

Stage 4.7 has a unique three-option approval gate:

  1. Approve -- Workflow complete. The full AI-DLC lifecycle is finished.
  2. Request Changes -- Provide revision feedback.
  3. Start New Ideation Cycle -- Feed the feedback-loop.md insights back into a new Stage 1.1.

This reflects the cyclical nature of the AI-DLC lifecycle.


Phase Summary

Deployment stages (4.1-4.3): - 4.1 Deployment Pipeline -- CD pipeline config, deployment strategy, rollback runbook - 4.2 Environment Provisioning -- AWS environment provisioning and validation with security posture review - 4.3 Deployment Execution -- Artifact deployment, smoke tests, health checks, database migrations

Operational readiness stages (4.4-4.6): - 4.4 Observability Setup -- Dashboards, alarms, SLOs, log queries, tracing, anomaly detection - 4.5 Incident Response -- Runbooks, incident plan, escalation matrix - 4.6 Performance Validation -- Load testing, NFR target validation, capacity planning

Continuous improvement (4.7): - 4.7 Feedback & Optimization -- SLO compliance, cost analysis, drift detection, feedback loop

Scope applicability: - enterprise / feature / workshop: All 7 stages - infra: Stages 4.1-4.4 (deployment-pipeline, environment-provisioning, deployment-execution, observability-setup) - security-patch: Stages 4.1, 4.3 (deployment-pipeline, deployment-execution) - mvp / poc / bugfix / refactor: Operation phase skipped entirely

Cross-References