> ## Documentation Index
> Fetch the complete documentation index at: https://invisly.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# How Invisly analyzes your code with AI

> Understand how Invisly analyzes your code using contextual AI models, pattern recognition, and multi-language support to deliver targeted assistance.

Once Invisly captures your screen, the analysis engine takes over. It does more than read text off an image — it interprets the full context of your coding problem, routes the analysis to the most appropriate AI model, and returns a response that addresses what you're actually trying to accomplish. This section explains how that process works and what types of problems you can bring to it.

## Capabilities

<AccordionGroup>
  <Accordion title="Contextual understanding" icon="brain" defaultOpen={true}>
    Invisly analyzes the entire problem context visible in your capture — not just a single line of code. This includes error messages, function signatures, surrounding logic, variable names, and any instructions or prompts visible on screen.

    The result is that the AI understands **what you're trying to accomplish**, not just what's literally on the screen. When you're debugging, it considers the broader control flow. When you're implementing a function, it factors in how the rest of the code uses it. This context-awareness is what separates targeted help from generic suggestions.
  </Accordion>

  <Accordion title="Multi-model processing" icon="cpu">
    Complex problems are routed through specialized AI models optimized for different categories of coding challenges. Invisly selects the appropriate model based on what it detects in your capture.

    * **Algorithm problems** are handled by models optimized for reasoning about time and space complexity
    * **Debugging tasks** route to models trained on error patterns and diagnostic reasoning
    * **Data structure questions** go to models with strong knowledge of memory layout and access patterns
    * **General syntax and best practices** use models focused on language conventions and readability

    You don't choose the model yourself — the system handles routing transparently so you always get the most relevant response.
  </Accordion>

  <Accordion title="Pattern recognition" icon="search">
    The analysis engine identifies common coding patterns and anti-patterns in your code and uses them to produce targeted suggestions. When it detects an inefficient approach — a nested loop where a hash map would work, a repeated computation that could be cached, an off-by-one error in an index — it flags it and explains a better alternative.

    Pattern recognition works at multiple levels:

    * **Structural patterns**: how functions, classes, and modules are organized
    * **Algorithmic patterns**: sorting strategies, traversal approaches, recursion vs. iteration
    * **Error patterns**: common mistake types associated with specific language features
    * **Style patterns**: deviations from widely accepted conventions in a given language
  </Accordion>

  <Accordion title="Language-agnostic support" icon="code">
    Invisly understands and provides assistance across a broad range of programming languages. You don't need to configure a language preference — the engine detects the language from the code itself.

    Supported languages include:

    * **Python** — including standard library usage, list comprehensions, decorators, and type hints
    * **Java** — classes, interfaces, generics, and common frameworks
    * **JavaScript** — ES6+ syntax, async patterns, DOM interaction, and Node.js
    * **C++** — pointers, memory management, STL, and template usage
    * And more, including TypeScript, Go, Rust, Ruby, Swift, and SQL

    Mixed-language captures — such as a code editor alongside a terminal with shell output — are handled gracefully, with each portion analyzed in its correct language context.
  </Accordion>
</AccordionGroup>

## What you can use it for

Invisly is designed to help with the full range of challenges that come up during technical assessments and coding problems. Here are the primary use cases:

<CardGroup cols={2}>
  <Card title="Debugging complex code" icon="bug">
    When you're stuck on a bug, share your code and any visible error output. Invisly analyzes the logic, identifies where the failure likely originates, and suggests a fix with an explanation.
  </Card>

  <Card title="Algorithm optimization" icon="trending-up">
    Paste your current implementation and ask for improvements. Invisly evaluates your approach for time and space complexity and suggests more efficient alternatives where they exist.
  </Card>

  <Card title="Syntax and best practices" icon="check-circle">
    Get guidance on proper coding conventions, idiomatic usage for the language you're working in, and common style guidelines — without having to leave your editor.
  </Card>

  <Card title="Concept explanation" icon="book-open">
    When you encounter an unfamiliar pattern or concept in the problem, Invisly can break it down into a clear, step-by-step explanation tailored to what's visible in your capture.
  </Card>
</CardGroup>