> For the complete documentation index, see [llms.txt](https://leviaprotocol.gitbook.io/leviaprotocol/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://leviaprotocol.gitbook.io/leviaprotocol/core-concepts/metacognition/metacognition-stream.md).

# Metacognition  Stream

The window into Levia's mind

### Understanding Levia's Thoughts

Think of the Metacognition Stream as a transparent window into Levia's cognitive processes. Like watching a master craftsman explain their work as they do it, this stream reveals the thought process, decisions, and actions in real-time.

### How It Works

When Levia processes a request, it shares its thinking:

* Initial understanding of the request
* Decision-making process
* Module and API selections
* Success or failure outcomes
* Alternative approaches considered

For example:

```
Analyze this sales data and create a visualization"

Metacognition Stream:
> Understanding request: Data analysis and visualization task
> Checking data format compatibility
> Selecting analysis module: DataProcessor
> API call: Processing raw data
> Considering visualization options: Line chart vs Bar graph
> Selected: Line chart (reason: time-series data)
> Generating visualization...
```

### Interactive Guidance

The stream isn't just for observation - users can interact with Levia's thinking:

* Suggest alternative approaches
* Point out overlooked factors
* Guide decision-making
* Share domain expertise

This creates a collaborative relationship where both Levia and users learn from each other, leading to better outcomes and continuous improvement.

{% embed url="<https://streamlit.io/>" %}