AI Model Configuration and API Keys

Mind Map Wizard leverages artificial intelligence to transform your text prompts into structured mind maps. This functionality is powered by the OpenRouter.ai platform, and for maximum privacy and control, Mind Map Wizard operates on a "Bring Your Own Key" (BYOK) model. This means you provide and manage your own OpenRouter API key.

OpenRouter API Key Management

To utilize the AI generation features, you will need an OpenRouter API key. This key authenticates your requests to the AI service.

Entering and Saving Your API Key

When you attempt to generate a mind map without a previously saved API key, a dedicated popup window (showApiKeyPopup in scripts/new.js) will appear, prompting you to enter your OpenRouter API key.

1. Obtain Key: Navigate to https://openrouter.ai/keys↗ to generate your personal API key.
2. Enter Key: Paste your API key into the input field provided in the popup.
3. Save & Generate: Click the "Save & Generate Mind Map" button. The application will then attempt to generate your mind map using the newly provided key.

Once entered, your API key is encrypted and stored locally in your browser for future use. The saveApiKey function (scripts/new.js) handles the encryption and storage process.

Managing Your Stored API Key

You can manage your saved API key at any time through the "API Key Management" interface (showApiKeyManagement in scripts/new.js). This interface allows you to:

• View/Update: See if a key is currently stored and replace it with a new one.
• Remove: Delete the stored API key from your browser.

Your API key is retrieved and decrypted using the getStoredApiKey function (scripts/new.js) whenever needed for API calls.

Local Storage and Encryption

As part of Mind Map Wizard's commitment to user privacy, your OpenRouter API key is never sent to or stored on Mind Map Wizard's servers. Instead, it is stored locally within your browser's localStorage and is encrypted for added security. For more detailed information on how your data is stored and protected, please refer to the [Local Data Storage and Privacy] page.

AI Model Selection

Mind Map Wizard provides a model selector to allow users to choose from various AI models available through OpenRouter.ai. This gives you flexibility in selecting the model that best suits your needs for mind map generation.

The model selector is initialized by the initializeModelSelector function (scripts/new.js).

How to Select a Model

1. Open Dropdown: Click on the model selection button, which typically displays the currently selected model name (e.g., "Google Gemini-2.5 Flash Lite"). This action triggers toggleDropdown (scripts/new.js).
2. Search for Models: A search input field will appear. You can type keywords here to filter the list of available models. The filterAndRenderModels function (scripts/new.js) updates the displayed options based on your search term.
3. Browse Options: The available models are listed as interactive buttons, with popular models prioritized at the top. Each option is rendered by renderModelOptions (scripts/new.js).
4. Select Model: Click on the desired model from the list. The selectModel function (scripts/new.js) will update the selectedModel variable, save your preference locally, and close the dropdown.
5. Current Model Display: The updateCurrentModelDisplay function (scripts/new.js) ensures that the name of the currently selected model is always visible on the main button.

The getSelectedModel function (scripts/new.js) is used internally to determine which AI model to call for mind map generation. By default, google/gemini-2.5-flash-lite is the pre-selected model.

System Prompt for AI Generation

The AI's ability to generate accurate and well-structured mind maps is guided by a specific "system prompt" that defines the desired output format and content characteristics. This prompt is crucial for ensuring consistency and quality in the generated mind maps.

The exact system prompt used is as follows:

Create a comprehensive, fact-rich mind map about: ${input}

Generate the mind map as Markdown text using the following structure:

# Matching Mind Map Title
## Branch 1
### Sub Branch A
### Sub Branch B
## Branch 2

**Formatting Requirements:**
- Each text element must be aligned to a specific hierarchical level using a new line plus the appropriate number of # symbols
- Aim for 2-3 levels of depth to keep the mind map scannable and not overwhelming
- For large enumerations (6+ items), combine related items into comma-separated lists within a single branch rather than creating excessive sub-branches

**Content Requirements:**
- Include **specific, concrete details and facts**, not just category labels
  - ❌ Bad: "## Education" 
  - ✅ Good: "## Education: PhD in Physics from MIT (2015)"
- Avoid generic structural sections like "Overview," "Introduction," or "Conclusion" – this is a mind map, not an essay
- If the topic contains extensive information, prioritize breadth over depth and consolidate where necessary
- Focus on the most relevant and interesting information that creates a useful knowledge structure
- Make the branches have different lengths for making the mind map visually more interesting.

**Output Format:**
Structure your response exactly like this:
{
	"markdown": "# Main Topic\\n\\n## Subtopic 1\\n- Point A\\n- Point B\\n\\n## Subtopic 2\\n- Point C\\n- Point D"
}

This prompt guides the AI to:

• Structure: Produce output in Markdown format, using # symbols to denote hierarchical levels.
• Depth: Maintain a balance between breadth and depth, typically 2-3 levels, to ensure readability.
• Detail: Emphasize specific, factual details rather than generic labels.
• Relevance: Focus on key information, avoiding introductory or concluding remarks common in essays.
• Format: Adhere to a strict JSON output format, ensuring the Markdown content is properly escaped.

By providing these clear instructions, Mind Map Wizard ensures that the AI generates high-quality, actionable mind maps.