The /instructions command helps the Kernaro Assist understand how your Enterprise Architect (EA) repository is structured. Instead of relying on generic Enterprise Architect knowledge, it analyzes your repository and any available MDG Technologies to create a Model instruction.
This Model instruction contains information such as:
- Element types used in your repository
- Valid connectors between elements
- Naming conventions
- Stereotypes and modeling standards
- Repository-specific best practices
Once the Model instruction is created, the AI uses it in future conversations to provide recommendations and generate models that follow your organization’s standards.
Why Use /instructions? #
By default, the AI assistant only understands general Enterprise Architect concepts. Every organization, however, follows its own modeling standards and conventions.
Running /instructions allows the AI to learn your repository’s rules, ensuring that its suggestions are consistent with your team’s modeling practices.
Benefits #
| Without /instructions | With /instructions |
|---|---|
| Uses generic EA knowledge | Learns your repository standards |
| May suggest incorrect stereotypes or connectors | Uses repository-specific stereotypes and relationships |
| Generic naming recommendations | Follows your organization’s naming conventions |
| Requires frequent corrections | Produces accurate results from the first interaction |
Prerequisites #
Before generating instructions, ensure the following:
| Requirement | Description |
|---|---|
| Enterprise Architect Repository | An EA repository must be open. |
| MDG Technology (Recommended) | Load your modeling technology such as ArchiMate, BPMN, or SysML. |
| Existing Model Content | Existing diagrams and elements help the AI learn your modeling patterns, although empty repositories are also supported. |
| MDG Availability | If no MDG is loaded, the assistant will guide you to import an MDG XML file, place one in the workspace, or specify the technology manually. |
How to Access #
You can generate repository instructions in either of the following ways:
- Type /instructions in the chat.
- If no Model instruction exists, the assistant may display a reminder in the chat background. Click the reminder or type /instructions to begin.
How It Works #
When you run /instructions, the AI performs the following steps:
| Step | Description |
|---|---|
| 1. Scan Repository | Analyzes the repository structure and available MDG technologies. |
| 2. Ask Questions | Requests clarification if additional information is required. |
| 3. Generate Model instruction | Builds a Modeling Model instruction based on your repository and modeling practices. |
| 4. Review | Presents the generated Model instruction for your review. |
| 5. Save | Stores the approved Model instruction in the shared EA database. |
| 6. Restart Chat | Start a new conversation, so the AI begins using the newly generated rules. |
Example #
Scenario #
Your organization uses ArchiMate and has the following modeling standards:
- Application Components must begin with APP-
- Business Processes must use the stereotype BusinessProcess
- Application Components connect to Business Processes using Serving relationships only
Without repository instructions, the AI might respond with:
Create an Application Component connected to a Business Process using an Association relationship.
Although technically valid in Enterprise Architect, this does not follow your organization’s standards.
After running /instructions, the AI learns your repository rules and responds with:
The AI will automatically understand “Create APP-CustomerPortal as an Application Component and connect it to Customer Onboarding using a Serving relationship.”
The generated model now follows your repository conventions automatically.
Updating Existing Instructions #
If a Model instruction already exists, simply run /instructions again.
The assistant will provide options to:
| Option | Description |
|---|---|
| Re-run Full Analysis | Rebuild the entire Model instruction by scanning the repository once again. |
| Update Specific Sections | Refresh selected portions of the existing Model instruction without rebuilding everything. |
Shared Across Users
The Model instruction is stored in the shared EA repository. Once created, every user connected to the same repository benefits from the same modeling standards.
Automatic Resume
If the instruction generation process is interrupted, the assistant resumes from the last completed step the next time /instructions are executed.
One-Time Setup
In most cases, you only need to generate instructions once. Update them whenever your modeling standards or MDG technologies are changed.
Subscribe to our Newsletter
Get product updates, feature tips, and integration insights in your inbox.
