AWS Database Blog
Conversational Oracle EBS operations with CloudWatch MCP and Kiro CLI
Imagine asking your Oracle E-Business Suite (Oracle EBS) environment a question in plain English—”Is the concurrent manager healthy?”—and getting an immediate, comprehensive answer that pulls data from your monitoring infrastructure, application logs, and system status checks. This is possible by combining Kiro CLI, an AI-powered command line assistant, with Amazon CloudWatch through the Model Context Protocol (MCP), a standard that enables AI assistants to securely access external data sources and tools.
MCP acts as the bridge between Kiro’s natural language interface and your Amazon CloudWatch monitoring data. Without it, AI assistants would need custom integrations for every data source. MCP provides a standardized way for Kiro to query metrics, retrieve logs, and check alarm states, translating your conversational questions into the appropriate CloudWatch API calls.
This integration creates a conversational operations layer for Oracle EBS on AWS, empowering your team to:
- Query system health conversationally: Ask “Show me application tier performance over the last 4 hours” instead of navigating multiple consoles
- Accelerate troubleshooting: Iterate through diagnostic questions rapidly without switching between tools
- Automate routine checks: Generate comprehensive health reports with a single command
- Reduce context switching: Access metrics, logs, and system status through one interface
The operational challenge that this addresses:
Oracle EBS operations teams manage complex, multi-tier environments spanning database instances, application servers, concurrent managers, and web tiers. Daily monitoring requires juggling the AWS Console, SQL*Plus, application logs, and custom scripts. Each diagnostic question means context switching, remembering specific command syntax, and manually correlating data across systems. During incident response, this friction slows down the critical discovery phase.
In this post, you learn how to implement conversational operations for Oracle E-Business Suite (Oracle EBS) on AWS by connecting Kiro CLI with your monitoring infrastructure through the MCP. We walk through the technical architecture that enables natural language queries to retrieve CloudWatch metrics, analyze logs, and execute operational commands. You see practical examples of generating comprehensive health reports, checking concurrent manager status, and automating routine tasks like service restarts—all through conversational commands. By the end, you have step-by-step guidance with code examples to set up this solution in your own Oracle EBS environment, transforming how your team monitors and manages complex multi-tier deployments.
Solution overview
The solution combines three key components to enable conversational Oracle EBS operations: Kiro CLI as the natural language interface, the CloudWatch MCP Server as the data bridge, and your existing AWS monitoring infrastructure. When you ask a question like “Is the concurrent manager healthy?”, Kiro translates your request into specific CloudWatch API calls through the MCP protocol, retrieves the relevant metrics and logs, and presents the results in a digestible format. This architecture preserves your existing monitoring setup while adding an intelligent conversational layer that accelerates troubleshooting and routine operations.The following diagram shows how these components interact within your AWS environment, illustrating the flow from natural language query to actionable operational insights.
Figure 1. Architecture diagram for Amazon CloudWatch MCP Server and Kiro CLI
Kiro CLI
Kiro CLI adds conversational access to your AWS monitoring data, working alongside your existing tools. Your AWS Command Line Interface (AWS CLI) scripts and Console dashboards continue handling production automation and systematic monitoring—they’re built for repeatability and precision. Kiro handles the other side: exploratory questions during troubleshooting. Ask ‘Show me concurrent manager status‘ and get formatted results without constructing queries, then immediately follow up with ‘What about the database tier?‘ Each tool serves its purpose.
For Oracle EBS administrators, this means asking questions like ‘show me the status of my concurrent managers‘ or ‘restart the application services on server EBSAPP01‘ without needing to recall the exact AWS Systems Manager (SSM) command syntax or instance IDs. Kiro handles the translation from intent to execution.
Another benefit becomes apparent with this. If a command fails, Kiro explains what went wrong and suggests how to fix it. This contextual guidance helps teams resolve issues faster while building their AWS expertise organically.By keeping everything in the terminal, Kiro helps reduce context switching overhead such as between browser tabs, documentation sites, and command line sessions. You stay in the flow, focused on solving problems rather than looking up syntax. For Oracle teams managing complex EBS environments, this workflow can help reduce the time spent on routine operations.
Figure 2. Kiro CLI prompt
CloudWatch Model Context Protocol (MCP) Server
The Model Context Protocol is an open standard that enables AI tools to access structured operational data. Think of it as a bridge between your monitoring infrastructure and AI assistants—allowing them to query metrics, analyze logs, and understand system state without custom integrations.The CloudWatch MCP Server implements this protocol specifically for Amazon CloudWatch. When you ask Kiro CLI about your Oracle EBS environment, the MCP server retrieves relevant telemetry data and presents it in a way that Kiro can interpret and act upon. This happens automatically in the background, giving you intelligent responses grounded in actual system data.For Oracle EBS operations, the CloudWatch MCP Server provides four key capabilities:
Intelligent alarm analysis: When alarms fire, the server doesn’t just tell you what’s alerting, it examines historical patterns, correlates related metrics, and suggests probable causes based on your specific environment’s behavior.
Log intelligence: Point it at your Oracle EBS log groups, and it surfaces anomalies, identifies error patterns, and highlights unusual activity within your specified timeframe. This accelerates log analysis by automatically correlating events and presenting relevant findings, letting you focus on diagnosis rather than search.
Metric context: Ever wonder what a specific CloudWatch metric actually measures or which statistic you should use? The server provides clear explanations of metric definitions and calculation methods.
Proactive recommendations: Based on your workload patterns and metric behavior, it suggests alarm configurations with appropriate thresholds and evaluation windows—taking the guesswork out of monitoring setup.
How the tools work together
When you enter a natural language query into Kiro CLI, it uses the CloudWatch MCP server’s tools to interact with your AWS environment. For Oracle EBS scenarios, this typically involves querying Amazon Elastic Compute Cloud (Amazon EC2) instances, retrieving CloudWatch metrics, analyzing log streams, and executing commands using AWS Systems Manager.
You can explore available capabilities by launching Kiro (enter kiro-cli in your terminal) and using /tools to see all available tools or /mcp to view loaded MCP servers. This transparency helps you understand what’s happening behind the scenes.
Figure 3. CLI Server Tools loaded in your client session
Figure 4. CloudWatch MCP server loaded in your client session
Alignment with AWS Well-Architected Framework
This integration of Kiro CLI with CloudWatch for Oracle EBS operations demonstrates strong alignment with AWS Well-Architected Framework principles. By combining conversational AI with robust monitoring capabilities, the solution enhances operational practices across multiple pillars while maintaining the security and governance controls essential for enterprise environments.
Using this example with Oracle EBS operations maps directly to several Well-Architected Framework principles:
Operational Excellence: Natural language interfaces lower the barrier to automation. When team members can express operational tasks conversationally, you reduce the risk of manual errors and streamlines the process to codify operational procedures. This democratizes AWS operations across your Oracle team, regardless of their AWS CLI proficiency.
Note, this accessibility requires careful governance. Kiro CLI implements a granular trust model where developers designate which commands require explicit approval. For privileged operations, organizations should configure approval workflows requiring human confirmation, while the solution respects existing AWS Identity and Access Manager (IAM) policies to make sure that conversational commands can’t exceed granted permissions.
Reliability: Faster incident response directly improves system reliability. By consolidating monitoring data and providing AI-assisted analysis, you can identify and resolve issues before they impact users. The ability to quickly correlate metrics across your Oracle database, application tier, and infrastructure layers helps you understand failure modes and implement appropriate safeguards.
Performance Efficiency: The solution adds conversational access to your Oracle EBS operational data, supporting faster iteration during investigations. Your team can ask follow-up questions in natural language—’What about the previous hour?’ or ‘Show me the same pattern for the database tier’—without reformulating queries or switching contexts. This accelerates the exploratory phase of troubleshooting, letting your team test hypotheses and correlate system behavior more rapidly while your existing monitoring infrastructure and automation scripts continue handling systematic oversight.
Cost Optimization: Operational efficiency translates to cost savings, helping streamline operations that traditionally take longer to perform manually, you’re optimizing your most expensive resource—skilled personnel. Additionally, better visibility into system behavior helps you right-size resources and avoid over-provisioning.
Security: The solution respects IAM boundaries and Systems Manager security controls. Operations are designed to work within your existing security model with the added benefit of natural language audit trails that clarify what actions were taken and why.
Getting started: Prerequisites
Important: Before implementing this solution in production environments, we strongly recommend thorough testing in non-production environments (development, test, or sandbox). Validate that the solution meets your organization’s security, compliance, and operational requirements. Establish clear governance policies around AI-assisted operations, including approval workflows for sensitive actions and comprehensive audit logging.
Before diving into the use cases, ensure that you have the following in place:
Oracle EBS environment: A running Oracle E-Business Suite deployment on AWS. If you’re setting up a new environment, refer to the Oracle E-Business Suite on AWS documentation for deployment best practices.
Observability foundation: Configure CloudWatch to collect metrics and logs from your Oracle EBS environment. CloudWatch Application Insights doesn’t currently support Oracle E-Business Suite because it’s designed for SAP, .NET, SQL Server, and other specific workloads. You will need to set up custom CloudWatch metrics, log streams, and alarms for your Oracle EBS components. This includes:
- CloudWatch Agent on EC2 instances to collect system and application metrics
- Custom log groups for Oracle Database alert logs, listener logs, and EBS application logs
- CloudWatch Logs Insights queries for log analysis
- CloudWatch alarms for critical metrics and thresholds
Kiro CLI installation: Install Kiro CLI on your local machine or bastion host following the Kiro documentation.
Deploying the MCP server on a local laptop without VPN or Direct Connect does present security considerations that should be addressed:
Security implications:
- The MCP server uses your local AWS credentials to access CloudWatch APIs over the public internet
- Log data and metrics transit through AWS’s public endpoints using TLS encryption
- Your laptop becomes a potential attack vector if compromised while holding active AWS sessions
Recommended security mitigations:
- Network security: Deploy the MCP server on a bastion host or jump server within your VPC, accessed through AWS Systems Manager Session Manager or VPN connection
- Credential management: Use temporary credentials with least-privilege IAM policies, avoiding long-lived access keys on local machines
- Alternative architecture: Consider running Kiro CLI and the MCP server on an EC2 instance within your private subnets, removing the need for public internet access to sensitive monitoring data
- Data classification: For highly sensitive environments, implement AWS PrivateLink endpoints for CloudWatch APIs to keep all traffic within the AWS Backbone
CloudWatch MCP Server: Set up the CloudWatch MCP Server by following the instructions in the AWS Labs CloudWatch MCP repository.
AWS CLI configuration: Verify that your AWS CLI is properly configured with credentials and default Region settings. See AWS CLI configuration guide for details.
Security considerations: Implement least-privilege IAM policies for your Oracle DBAs and infrastructure engineers. Review Kiro CLI security best practices and Systems Manager security guidelines to ensure that your setup follows AWS security recommendations. For Oracle database connectivity, configure Kiro to use read-only database credentials to prevent the AI assistant from inadvertently executing disruptive SQL statements such as DELETE, UPDATE, INSERT, or DDL commands. This acts as a safety net so that conversational queries remain non-destructive regardless of how requests are interpreted.
Configuring Kiro CLI steering for Oracle EBS
Why steering files matter
When you interact with Kiro CLI without steering files, it needs to discover your environment details with every request—asking questions, making assumptions, or executing exploratory commands to understand your Oracle EBS setup. This discovery process consumes time and can lead to errors if assumptions are incorrect.
Steering files solve this problem by giving Kiro persistent knowledge about your Oracle EBS environment. Think of them as a reference guide that Kiro consults before taking action. With steering files in place, Kiro immediately knows your instance naming patterns, which OS users to use for different operations, where to find logs and scripts, and how your environments are structured. This removes the discovery phase and allows Kiro to execute operations accurately and efficiently from the first request.
We recommend that you let Kiro CLI help you create steering files. You can start an interactive session and provide the template structure you want:
Kiro will create the file directly in the .kiro/steering/ directory and may automatically populate it with information from your environment (such as EC2 instance details if it can discover them). Review and customize the generated file to ensure that all values match your specific setup.
The steering file will be in markdown format and stored at:
Example Oracle EBS environment steering file
Important: The following is an example steering file for reference. You must customize all values to match your specific Oracle EBS environment. Pay special attention to: – OS usernames (applmgr and oracle, and the Secrets Manager secret in this example—replace with your actual values)
The file .kiro/steering/oracle-ebs-environment.md will be created with content similar to this example:
- Learn More: AWS Secrets Manager Documentation
- Creating Secrets: Store a new secret in Secrets Manager
Resource tagging strategy
Beyond environment configuration, a well-designed tagging strategy is essential for Kiro to automatically identify and operate on the correct Oracle EBS resources. Consistent tagging enables Kiro to translate natural language requests like “check the TEST database servers” into precise AWS API calls that target the right instances. For comprehensive guidance on implementing tagging best practices, see the AWS Tagging Strategies documentation.
Note: Adapt these tags to match your organization’s tagging standards.
Oracle EBS resources use the following tags:
- Application: OracleEBS
- Environment: Development | Test | Production
- ApplicationRole: Database | AppServer | ConcurrentManager (comma-separated for multiple roles)
- EBSInstance: DEV | TEST | PROD
- EBSVersion: 12.2.11
- Tier: Database | Application
- Owner: OracleDBTeam (replace with your team name)
- CostCenter: Finance-IT (replace with your cost center)
Example Tagging for Oracle EBS Resources:
Database Server:
Application Server:
How Kiro uses tags
When you ask Kiro to “check the status of development application servers” or “restart services on the TEST concurrent manager”, it uses these tags to identify the correct resources. For example, a query like “show me all Oracle EBS test database instances” translates to an AWS CLI command filtering on tags like Application=OracleEBS, Environment=Test, and ApplicationRole=Database.
This tag-based approach provides several benefits:
- Simplified operations: You reference resources by their logical function rather than technical identifiers
- Reduced errors: Clear tagging minimizes the risk of operating on the wrong resources
- Better automation: Consistent tags enable reliable automation across your Oracle EBS landscape
- Improved visibility: Tags facilitate cost tracking, compliance reporting, and resource inventory management
Use Case 1: Automated health reporting for Oracle EBS
One of the most time-consuming tasks for Oracle administrators is generating comprehensive health reports. Traditionally, this involves connecting by using SSH (Secure Shell) into multiple servers, running various status commands, checking log files, and compiling everything into a coherent report. With Kiro CLI and CloudWatch MCP, you can generate detailed health assessments with a single natural language request.
In this scenario, we’ll ask Kiro to assess the health of our Oracle EBS test environment, including the database, application servers, and concurrent managers. Behind the scenes, Kiro will:
- Identify relevant EC2 instances using tags
- Query Oracle Database status via Systems Manager
- Check Concurrent Manager health and workload
- Examine Application Server processes and resource utilization
- Analyze recent log entries for errors or warnings
- Compile everything into a structured health report
Note: The following examples are condensed for readability. In practice, Kiro provides more detailed output.
Example: Comprehensive environment health check
Your request: Generate a health report for the Test Oracle E-Business Suite database, application servers, and concurrent managers.
What Kiro does behind the scenes:
When you submit this request, Kiro CLI orchestrates multiple operations automatically:
- Discovers infrastructure: Queries EC2 to identify database and application server instances using tags
- Checks AWS health: Verifies instance status, system status, and EBS volume health
- Gathers instance details: Retrieves instance type, IP addresses, launch times, and uptime
- Executes remote commands: Uses SSM to run health check scripts on both servers
- Analyzes results: Processes output from database processes, concurrent managers, disk usage, and memory utilization
- Generates report: Compiles all data into a structured health report with recommendations
Kiro’s response summary:
Key sections from the generated report:
Executive summary: – Overall Status: HEALTHY – Database Server: Running – Application Server: Running – Concurrent Managers: 6 Active
Infrastructure health: – Database Server (EBS-TEST-DB-01): r6i.xlarge, 131 days uptime – Application Server (EBS-TEST-APP-01): r6i.4xlarge, 111 days uptime – AWS Health checks passing
Database health: – Oracle Database (VISCDB): Running – Listener Status: Running – Process monitoring confirms healthy operation
Application tier health: – Application Listener: Running – Concurrent Processing: 6 active FNDLIBR processes – Service Manager (FNDSM): Running
Recommendations: – High Priority: Address storage capacity warnings (82–86% utilization) – Medium Priority: Configure swap space, enable CloudWatch monitoring – Low Priority: Schedule maintenance window for patching
Full report: View complete health report output (link to full output)
What this demonstrates:
This example shows how Kiro CLI removes the manual work of: – Connect by using SSH into multiple servers, running individual status commands, checking process lists and resource utilization, compiling data into a coherent report, and Identifying issues and generating recommendations.Instead, a single natural language request produces a comprehensive, actionable health quickly and automatically.
Example: Concurrent Manager status check
Your request: What is the current status of Oracle Concurrent Managers for the TEST environment?
What Kiro Does Behind the Scenes:
- Identifies target instance: Uses tags to find the TEST environment application server
- Executes remote command: Sends SSM command to check for FNDLIBR and FNDSM processes
- Retrieves results: Waits for command completion and fetches output
- Analyzes process list: Parses process information including PIDs and CPU time
- Presents summary: Formats the data into a clear status report
Kiro’s response:
Full output: View complete concurrent manager status output (link to full output)
When you need detailed information about specific components, Kiro can drill down into individual subsystems. This multi-step orchestration, discovering instances, executing remote commands, and synthesizing results, happens automatically based on your natural language request.
Use case 2: Streamlining routine operations
Beyond reporting, Kiro CLI excels at executing operational tasks. Oracle administrators perform numerous routine activities: checking service status, restarting components, verifying configurations. These tasks are straightforward but time-consuming, especially when you need to look up instance IDs, remember exact command syntax, or wait for remote commands to complete.With Kiro, you describe what you want to accomplish, and it handles the orchestration. Let’s look at a common scenario: checking application server status and performing a controlled restart.
Example: Application Server status check
Your request: Get status of Oracle EBS application server for Test. Do not restart the server, only report on current status.
What Kiro does behind the scenes:
- Identifies target server: Locates the TEST application server using tags
- Executes comprehensive check: Runs multiple status commands via SSM to check all components
- Gathers component status: Checks concurrent managers, Apache/OHS, Forms server, OACore, and TNS listener
- Collects system metrics: Retrieves uptime, load average, and memory utilization
- Analyzes results: Processes all output and identifies potential issues
- Presents structured report: Formats data into clear component status with visual indicators
Kiro’s response summary:
Full output: View complete application server status output (link to full output)
Kiro automatically locates the correct instance, executes the appropriate Oracle EBS status commands, and presents the information in a structured format. You don’t have to remember the instance ID, SSM command syntax, or Oracle EBS script locations. The visual indicators (✅ and ❌) make it immediately clear which components are healthy and which need attention.
Example: Controlled service restart
Your request: Restart Oracle EBS application services on EBS-TEST-APP-01.
What Kiro does behind the scenes:
- Acknowledges critical operation: Recognizes this is a downtime-causing operation requiring careful execution
- Retrieves credentials: Securely fetches APPS password from AWS Secrets Manager
- Executes stop sequence: Runs adstpall.sh to gracefully stop all application services
- Waits for clean shutdown: Allows 30 seconds for services to stop completely
- Executes start sequence: Runs adstrtal.sh to start all application services in proper order
- Monitors progress: Polls command status and waits for completion (3–5 minutes)
- Verifies services: Checks that all components are running after restart
- Reports status: Provides confirmation with process counts for each component
Kiro’s response summary:
Full output: View complete service restart output (link to full output)
Here, Kiro orchestrates a multi-step process: stopping services in the correct order, waiting for clean shutdown, starting services in the proper sequence, and verifying that everything came back up successfully. This level of automation reduces the risk of human error during maintenance windows while providing clear visibility into each step. The automatic credential retrieval from AWS Secrets Manager supports secure password handling without manual intervention.
Key takeaways
The combination of CloudWatch MCP Server and Kiro CLI fundamentally changes how Oracle EBS teams interact with their AWS infrastructure. By consolidating metrics, logs, and operational commands into a conversational interface, you eliminate the friction that slows down daily operations.
Important: Understanding AI-assisted operations
While conversational interfaces streamlines operations, AI agents like Kiro are non-deterministic and might interpret requests differently across interactions—always validate proposed commands before execution and require human approval for privileged operations. As noted in the Prerequisites section, thorough testing in non-production environments and clear governance policies are essential before production deployment.
Your team gains a unified view of the entire Oracle EBS stack—from database performance to application tier health to concurrent processing workload. When issues arise, you can quickly correlate data across these layers without jumping between tools. For routine operations, natural language commands replace the need to memorize syntax or maintain runbook documentation.
This approach aligns with AWS operational excellence principles by making automation accessible, reducing manual errors, and accelerating incident response. As your team becomes more comfortable with these tools, you will find new ways to use them for capacity planning, change management, and proactive monitoring.
What’s next?
In future posts, we will explore advanced scenarios including:
- Maintenance planning workflows that coordinate changes across your Oracle EBS landscape
- Deep-dive root cause analysis techniques using historical metric correlation
- Cost optimization strategies informed by operational patterns
We encourage you to experiment with this solution in your non-production environments first. Start with simple queries and gradually build up to more complex operational workflows. Validate the solution thoroughly before considering production use, ensuring it meets your organization’s security, compliance, and operational requirements.
Learn more
- Oracle E-Business Suite on AWS
- AWS Labs CloudWatch MCP Server
- AWS Systems Manager Documentation
- Amazon CloudWatch Documentation
In this post, we demonstrated how AI-assisted operations can enhance Oracle E-Business Suite management on AWS, providing Oracle administrators and infrastructure teams with powerful new capabilities for monitoring, troubleshooting, and maintaining their environments.