InfraNodus Graph View VSCode / Cursor AI Extension

This VSCode / Cursor AI / Antigravity extension adds a graph view to the Secondary Side Bar (or wherever else you prefer in your IDE) that displays the InfraNodus graph visualization.

It works on both text and code — visualize the topics and connections in markdown files, or map out the architecture of a codebase (functions, classes, variables, and the references between them) to find connections, identify the main and latent components, and reveal the gaps.

This is super useful for:

• Exploring and optimizing your knowledge base and LLM wiki
• Steering your LLM's reasoning
• Finding blind spots in your AI agents' docs, PRDs, and rules
• Exploring and optimizing your code base

Features

• Displays InfraNodus graph in VSCode's Secondary Side Bar (or in Cursor's AI agent bar) — see the help section below to understand how to put it into the right sidebar.

• Use it on individual files or folders (right-click to activate)

• Embedded web view with the InfraNodus graph interface — Obsidian-like but better, because it includes network stats and analysis

• AI-powered topic modeling of your content

• Code architecture graphs: visualize how functions, classes, methods, and variables in your codebase reference each other — click any node to jump to its definition

• Auto-detects whether to run plain text analysis or code-architecture analysis based on file extension (configurable)

• Can visualize the diff between files / folders / project

• Use the graph interface to search for relevant content and topics

• Can be used to detect gaps between content blocks

• Click the AI buttons to generate prompts and paste it into your LLM chat (Claude code, Cursor AI, etc.)

• Has an InfraNodus Log view (which you can put next to Terminal) with compressed graph / selection data that you can paste to your LLM (via Claude Code or Cursor) in order to improve the quality of the output.

Requirements

• You can use this extension with VSCode 1.84.0, Antigravity IDE, Windsurf AI, Cursor AI

• You can open your Obsidian vault in any of the editors above to get advanced AI capabilities without losing the graph view capacity

• You need an InfraNodus account to use this extension. You can sign up for a free trial at https://infranodus.com and then obtain the key at the InfraNodus API Access Page.

Use Cases

This extension can be super useful for the following scenarios:

Obsidian: Gap Analysis & Knowledge Optimization

Use it with your Obsidian vault to optimize your content structure or to in the Karpathy's LLM Wiki setup.

This extension's Gap Analysis will show you the clusters of ideas that are not well connected and generate prompts and/or AI ideas that will help you develop this content further.

The extension is based on the InfraNodus cognitive variability framework that optimizes your discourse based on its structure. If your ideas are too connected, it will propose to develop latent topical clusters and peripheral ideas. If they are too dispersed, it will help you make your discourse more coherent.

Steer Your Model's Reasoning

Another powerful use case is to use the graph to steer your model's reasoning using the prompts it generates. For instance, you can select the clusters that are not linked yet (content gap) and generate an AI prompt that helps the model produce a response that links those clusters and bridges the gap — whether they're disconnected topics in your notes or two unrelated parts of your code.

Optimize Your AI Agent's Docs, Skills, Rules, and History

You can also point the extension to the .md or .mdc files generated by your coding agents, such as Cursor AI or Claude (available in ~/.claude or ~/.cursor folders in your home or project folder). You can then analyze an auto-generated doc or PRD (Product Requirements Document) to reveal the main ideas and potential gaps. Then use these gaps to address the blind spots and optimize your agents' reasoning and workflows.

Explore Your Codebase

You can use the extension to explore an unfamiliar codebase — getting a holistic view of the main clusters and ideas, and revealing the blind spots that you can bridge with new ideas (or new functions).

When you point the extension at a code file or folder, it builds an architecture reference graph: nodes are functions, classes, methods, and exported variables; edges are containment and references. Clicking a node in the graph jumps the editor straight to that symbol's definition. This requires a language server for the file (TypeScript/JavaScript, Python, Go, etc.) — VSCode and its forks ship most of these out of the box. The mode is picked automatically per file (text extraction for .md, .txt, .rst, …; code-architecture graph for .ts, .py, .go, …), and you can override it in Settings → InfraNodus: Text & Code Analysis.

Notes

• This extension is a work in progress and is currently in alpha. We are working on adding more features and optimizing the user experience. For feedback, please, open an issue on Github.

• For prose (markdown, plain text, docs), the extension shines on Obsidian vaults and LLM wiki setups.

• For code, the extension builds an architecture reference graph — nodes are symbols (functions, classes, methods, exported variables) and edges are containment and references resolved via the language server. It's ideal for getting a bird's-eye view of an unfamiliar folder, spotting tightly-coupled clusters, and finding orphaned or weakly-connected symbols. Code-mode results depend on the language server's coverage of the file, so language support tracks what VSCode itself supports.

How to Use

0. Install the extension via the Extensions marketplace or manually (see instructions below for manual installation)
1. Open VSCode's Secondary Side Bar (View -> Secondary Side Bar)
2. Look for "InfraNodus Graph" view
3. Get the API key at https://infranodus.com/api-access and add it using the Key button in the graph window (bottom left) or using Cmd + Shift + P > InfraNodus API key menu in IDE's preferences
4. You might want to move the InfraNodus graph view elsewhere. Right click the extension's icon and choose the new location for it.
5. The graph should load automatically in the view for the currently active file. Click the Reload button if the graph doesn't load.
6. Right-click on a file or folder to open it in the InfraNodus Graph view
7. Open the InfraNodus Log in terminal (using commands Cmd+Shift+P -> InfraNodus Graph: Open InfraNodus Log)
8. Use commands (Cmd+Shift+P -> InfraNodus Graph: Paste (Selected) Graph in a File) to copy the (selected) graph data to a file — this is useful for using with AI co-pilot chatbots
9. Use the features outlined above to navigate and search through your content using the graph
10. Use the AI buttons to generate a prompt in the InfraNodus Log view and then copy it to your favorite LLM chat (e.g. Claude Code, GitHub Copilot, Codeium, Continue, Antigravity, Cascade in Windsurf AI)
11. You can customize the prompts generated when you click on the graph's AI and context buttons in the extension's settings.
12. You can also customize the type of processing. For instance, by default the extension will prioritize [[wiki links]] and #hashtags to words, but will also include single words if there are no [[wiki links]] in your text. You might want to choose to process [[wiki links]] only in the settings.
13. You can add "stopwords" in the settings: the words and terms that the graph should not process. Useful for auxiliary terms and function names.
14. By default, the extension uses Auto mode for content processing: prose files (.md, .txt, .rst, …) are parsed as text, and code files (.ts, .py, .go, …) are parsed as a code-architecture graph where nodes are functions/classes/variables and clicking a node jumps to its definition. You can override this in Settings → InfraNodus: Text & Code Analysis → Content To Send (e.g. force "Parsed Text Only", "Parsed Code", or "Full File Contents").

How to Position the Extension in the Right Sidebar

By default, the extension is positioned on the left, next to the Explorer panel. You might want to move it to the right sidebar.

In order to do that, press the key combination Cmd/Ctr + Shift + P to open the Command Palette.

Then choose View: Move View

Select the InfraNodus Graph

Then pick Secondary Side Bar

The extension will be moved to the secondary side bar on the right where the AI chats are positioned. This is usually a more convenient place for it.

Full video with a demo

Recommended Tools

• You will need an InfraNodus account to use this extension
• We also recommend to install the InfraNodus MCP Server to your IDE as your agents can then automatically generate and retrieve graph insights directly during your LLM interactions.
• This extension can still be useful as you can see what's happening under the hood and, in some instances, use a visual interface to steer your LLM's reasoning

Manual Installation

1. Get the InfraNodus Graph View .vsix file from the releases page
2. Open VS Code
3. Press Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (Mac)
4. Type Extensions: Install from VSIX
5. Select the .vsix file you downloaded
6. Press Enter
7. Verify installation by clicking the Extensions icon in the sidebar and checking the InfraNodus Graph View extension
8. Activate the extension (if it hasn't been done already)
9. Add your InfraNodus API key (using commands Cmd+Shift+P -> InfraNodus Graph: Set API Key or using the Key button at the top left of the InfraNodus Graph View.
10. You might need to reload the extension or your VSCode

Installation for Developers

1. Clone this repository
2. Run npm install to install dependencies
3. Run vsce package to build the extension
4. This will create a .vsix file in your project directory
5. Open VS Code
6. Follow the steps in the Manual Installation section above

To then publish the extension:

1. Update version in package.json
2. Run vsce package
3. Copy the new .vsix file to the repo that is to be published
4. Commit both repos
5. Go to the repo that is to be published
6. Run git tag 0.x.x
7. Run git publish origin 0.x.x
8. Run vsce logout your_user_name
9. Run vsce publish

Updates

1. Check the releases page for new versions
2. Reinstall the extension
3. Add your InfraNodus API key again