Get started and Chat Behaviour
Getting Started
Follow these steps after Setup on Kernaro AI for EA right away:
Open the Chat
- Go to Add-Ins > Kernaro AI (Beta) > Chat.
- A chat panel will appear inside EA.
Start Asking Questions!
Try these example prompts to get started:
- “What models and packages are in this repository?”
- “List all packages in the project”
- “Show me the elements in the current diagram”
Chat Behaviour
This section explains how chat operates within Kernaro AI and highlights important behaviours and controls.
Repository-Specific Instructions (/instructions)
When the chat window is empty, the /instructions option is available. Typing /instructions allows Kernaro AI to analyze your current EA repository and generate repository-specific guidelines such as:
- Naming conventions
- Element types
- Package structures
Once generated, these guidelines are automatically applied to all future chat sessions within that repository.
What /instructions does
Kernaro AI examines your repository – the element types, naming patterns, package structures, and modeling frameworks you’re using – and produces a set of guidelines covering:
- What types of elements and connectors are used in your project
- How things are typically named
- How packages are organized
- What tags and properties are commonly used
These guidelines are automatically loaded at the start of every future chat session. After running this once, Kernaro AI will follow your repository conventions consistently.
Best Practice
- Run /instructions when starting with a new project
- Re-run it whenever your modeling standards change
Stopping a Chat Response
While the AI is generating a response or executing actions, a Stop button is displayed.
- Click Stop to cancel the current operation
- Any actions already completed before stopping will not be reversed
Navigation Restrictions During Streaming
While the AI is actively streaming a response, navigation to the Agents Executions tabs in the Setup window is temporarily restricted. Wait for the response to complete (or click Stop) before switching tabs.
Thinking Mode
A lightbulb icon is available near the chat mode selector.
- Click to enable or disable Thinking Mode
- When enabled, the AI displays step-by-step reasoning before the final response
- Useful for complex or multi-step tasks
- Can be toggled at any time without restarting the conversation
Event-Mapped Agents and Chat
If you have an agent mapped to an EA event (e.g., “element created”), that agent is triggered only by direct user actions in EA – not by changes the chat AI makes. For example, if the chat creates a new element as part of your request, the “element created” agent will NOT fire for that element. Event-mapped agents respond to user-initiated EA actions only.
Modeling Guidelines
Use the /instructions command to have Kernaro AI analyze your repository and create modeling guidelines automatically. These guidelines are stored per-repository and loaded at the start of each chat session, ensuring the AI follows your project’s naming conventions, element types, and package structures consistently.
Troubleshooting – Chat
Chat is not responding
- Make sure you have at least one AI model set up and marked as Chat Default.
- Try closing and reopening the Chat window from Add-Ins > Kernaro AI (Beta) > Chat.
- Check your internet connection.
Navigation is blocked while AI responds
- This is expected behaviour during streaming. Wait for the response to finish, or click Stop to cancel it.
Frequently Asked Questions – Chat
Can I use different models for different chats?
Chat always uses the default model set in Settings. To use a different model, update the Default Chat Model in Setup > Settings.
Will the AI remember my previous conversations?
Each chat session starts fresh. Use /instructions to persist repository-specific guidelines across sessions.
Can I stop the AI mid-response?
Yes. Click the Stop button that appears while the AI is generating. Any actions already completed will not be reversed.
