In Claude models, most times, the discussion/steps/points are specified with nice emojis like checkboxes, warnings, etc. But I don’t see that with GPT5, is there a way to make GPT5 to show Emojis? Is there a way to enable them for GPT5 as well? It will make reading the summary, discussion, etc. much easier.
Overall I feel GPT5 is better with coding in complex codebases, so I want to stick with GPT5 model.
Make a rule and test that?
Hi sidscrazy,
You’ve hit on a great point. Using emojis for structure makes AI responses much easier to read. While Claude models often do this by default, you can definitely instruct GPT-5.
Since you want to stick with the GPT model for its strength in complex codebases, here is a set of instructions on how you can enforce a consistent, emoji-rich output.
Most AI environments, including Cursor, allow you to set custom instructions or “rules” that are applied to every conversation. You can usually find this in your settings under a name like “AI Rules,” “Custom Instructions,” or by editing the AI prompt configuration for your workspace.
Here’s how to do it:
Find the settings panel in Cursor for custom AI instructions.
Add a new rule or edit the existing instructions.
Paste the following text into the instructions box. This will tell the AI how you want it to behave in every response.
Instruction to Paste:
Always use emojis to structure your responses for better readability. You must follow this convention strictly:
- 💡 For ideas, suggestions, or key insights.
- ✅ For completed steps, key takeaways, or positive confirmations.
- ⚠️ For warnings, potential pitfalls, or important notes.
- ❌ For errors, anti-patterns, or things to avoid.
- 📝 For summaries, plans, or detailed explanations.
- ⚙️ For technical details, code, or configuration instructions.
- ➡️ For next steps or calls to action.
When presenting lists, discussions, or multi-step processes, use these emojis to introduce each point. This is a mandatory formatting rule for all your outputs.
By setting this rule, you only have to define it once, and the GPT model will follow these formatting instructions in all future conversations.
Creating a markdown file is an excellent and very flexible alternative. This is a great way to keep your rules version-controlled with your project.
Here’s the workflow:
Create a file in the root of your project named ai-rules.md.
Paste the rules into this file. You can use the same list from Option 1.
Content for ai-rules.md:
# AI Formatting and Emoji Rules
As an AI assistant, you must adhere to the following formatting rules for all responses in this project.
## Emoji Conventions
- 💡 **Idea/Suggestion:** Use for new ideas or suggestions.
- ✅ **Done/Takeaway:** Use for completed items or key results.
- ⚠️ **Warning:** Use for important warnings or potential issues.
- ❌ **Error/Avoid:** Use for explaining errors or things to avoid.
- 📝 **Summary/Plan:** Use for summaries, lists, and plans.
- ⚙️ **Technical Detail:** Use for code snippets or configuration.
- ➡️ **Next Step:** Use for actionable next steps.
Instruct the AI: At the beginning of a chat or for a specific complex task, you can now simply tell the AI:
“Use the formatting guidelines in ai-rules.md to structure your entire response.”
The most powerful solution is to use a custom rule (Option 1) to make the markdown file (Option 2) the permanent source of truth.
Set your custom AI instruction in Cursor to this one simple line:
“For every response, you must strictly adhere to the formatting conventions and emoji usage defined in the ‘ai-rules.md’ file located in the project root.”
This way, you get the consistency you want from the GPT model automatically, without having to remind it every time.
Have fun with Cursor IDE!