### Constraining the Model with Competency Prompts
To get the best results, it is essential to constrain the AI model by giving it precise instructions and good practices. This is called a **competency prompt**: a reference document (style guide, methodology, business rules) that you attach to your requests to steer generation.

Concrete examples:
- **Vibe Coding (HTML5 block)**: when using AI generation to create interfaces via the HTML5 block, attach your HTML/CSS style guide to the prompt. The model will then respect your colours, typography and components.
- **Markdown document generation**: attach a documentation methodology in Markdown (standard structure, heading conventions, glossary) so the AI produces documents that match your standards.

Thanks to the text-variable attachment system, you can store your style guides and methodologies as project variables and inject them into every AI conversation with one click.
### Major Use Case: Vibe Coding with the HTML5 Block
One of Celestory's most powerful use cases is vibe coding:
1. Open the AI chat in the Text tab.
1. Attach your HTML/CSS style guide as a text variable.
1. Describe in natural language the interface, animation or component you want.
1. Copy the generated code into an HTML5 block in your graph — or, with a tool-capable model, ask the assistant to write it straight into the block.
1. See the result instantly in the nocode engine.
1. Iterate through AI conversation until you get what you want — without writing a single line of code by hand.
### Monitoring tokens and credits
Celestory displays in real time:
- **History tokens**: how many tokens your conversation has already used against the model's context window.
- **Input tokens**: how many tokens your next message will add (attached variables included).
- **Model cost**: the `©` badge next to the model picker.
- **Monthly quota**: the "Monthly usage" gauge at the top of every tab.

Watch these to optimize your prompts, pick the model that fits your quality/cost ratio, and anticipate quota overruns before a production launch.
### What about "advanced mode"?
An **Advanced** switch exists in the Image, Video and Audio tabs: it replaces the whole form with a free-text model name and a JSON parameters area.

🔺 **It is only visible to Celestory administration accounts**, not to an ordinary creator account. If you cannot see it, that is not a bug: use the model picker and the fields offered.

---
→ Further reading: [The AI Assistant That Builds Your Graph](Assistant_Graphe.md) · [Plugging an MCP Server into the AI Assistant](Connecteurs_MCP.md) · [The AI Text Request Block: What It Won't Tell You](Bloc_Requete_IA_Pieges.md) · [AI Generation — Player Side](AI%20Generation%20Player%20v2.md) · [AI Glossary](AI%20Glossary%20v2.md) · [AI Credits and Subscription](../../../Découvrir%20Celestory/Découvrir%20Celestory%20v3/EN/Credits_IA_Et_Abonnement.md)