# Understand Anything

Source: https://trendhub.net/skills/egonex-ai-understand-anything
Repository: https://github.com/Egonex-AI/Understand-Anything
Author: EgonexAI (https://github.com/Egonex-AI)
License: MIT
GitHub stars: 92,877

Understand Anything is a plugin and skill set that has your coding agent build a knowledge graph of a codebase, wiki or Figma file, then answer questions, explain components, assess diffs and write onboarding guides from that graph. An interactive dashboard shows layers, guided tours and business flows. It suits developers joining unfamiliar or large projects.

## What problems it solves

- **A large unfamiliar codebase, nowhere to start**: A multi-agent scan maps files, functions, classes and dependencies into a graph with architectural layers and a guided tour ordered by dependency.
- **Agents reread the same code every session**: Chat, explain and onboarding skills grep a saved graph for relevant nodes and edges instead of loading whole source trees.
- **Change impact is hard to judge**: The diff skill maps changed files to graph nodes, follows one hop of edges and rates risk by complexity and layers crossed.
- **Business logic hides behind technical structure**: A domain skill extracts domains, flows and steps and shows them as a separate horizontal flow view.

## Install

### Claude Code

Installs the native plugin, including its analysis subagents and hooks.

```
/plugin marketplace add Egonex-AI/Understand-Anything
/plugin install understand-anything
```

### Codex, Gemini CLI, OpenCode, Cline, Kiro, Trae and other agents (macOS/Linux)

Downloads a copy into ~/.understand-anything/repo and links the skills for the chosen platform; in Codex, invoke skills with $ instead of /.

```
curl -fsSL https://raw.githubusercontent.com/Egonex-AI/Understand-Anything/main/install.sh | bash
curl -fsSL https://raw.githubusercontent.com/Egonex-AI/Understand-Anything/main/install.sh | bash -s codex
```

### Windows (PowerShell)

The same installer for Windows, with its own update and uninstall switches.

```
iwr -useb https://raw.githubusercontent.com/Egonex-AI/Understand-Anything/main/install.ps1 | iex
```

### VS Code + GitHub Copilot

Copilot in VS Code 1.108 or later can also pick the plugin up automatically from a cloned repo.

```
/plugin marketplace add Egonex-AI/Understand-Anything
/plugin install understand-anything@understand-anything
```

### Copilot CLI

Installs the plugin folder directly from GitHub.

```
copilot plugin install Egonex-AI/Understand-Anything:understand-anything-plugin
```

### Devin

The plugin route gives namespaced skills and auto-update hooks; an optional MCP server can serve existing graphs.

```
devin auth login # plugin management requires sign-in
devin plugins install Egonex-AI/Understand-Anything#understand-anything-plugin
```

Claude Code users install through the plugin marketplace, other agents through the one-line installer; the analysis scripts need Node.js 22 or later and pnpm 10 or later. The first /understand run on a large project can use many tokens, so the README suggests a subscription plan or a local model, and later runs only re-analyse changed files.

## Skills

Skill names are shown as /name. Depending on the agent, a skill runs when it is relevant or when you call it by name; plugin installs may add a prefix.

- `/understand-chat`: Answers questions about a codebase by grepping the graph for matching nodes and following their edges. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-chat)
- `/understand-dashboard`: Starts the local web dashboard that renders the saved graph by layer, with search and node details. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-dashboard)
- `/understand-diff`: Maps current changes or a pull request to graph nodes, lists affected components and layers, and rates the risk. (https://trendhub.net/skills/egonex-ai-understand-anything/understand-diff.markdown)
- `/understand-domain`: Pulls business domains, flows and process steps out of the code and shows them as a horizontal flow graph. (https://trendhub.net/skills/egonex-ai-understand-anything/understand-domain.markdown)
- `/understand-explain`: Explains one file, function or module in depth, using the graph to show what it depends on and what calls it. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-explain)
- `/understand-figma`: Reads a Figma file through its REST API and graphs pages, screens, components, instances and design tokens. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-figma)
- `/understand-knowledge`: Turns a Karpathy-style markdown wiki into a graph of articles, entities, claims and topic clusters. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-knowledge)
- `/understand-onboard`: Writes an onboarding guide for new team members from the graph's layers, tour and complexity hotspots. (https://github.com/Egonex-AI/Understand-Anything/tree/main/understand-anything-plugin/skills/understand-onboard)
- `/understand`: Scans a codebase with parallel subagents and saves a knowledge graph of files, symbols, layers, a guided tour and per-file lessons. (https://trendhub.net/skills/egonex-ai-understand-anything/understand.markdown)

## Questions

### Is it free, and what does it cost to run?

The code is MIT-licensed. The cost is model tokens: the first full analysis of a large project can be expensive, while later runs only re-analyse changed files.

### Which agents are supported?

Claude Code natively, plus Cursor, VS Code with Copilot, Copilot CLI, Codex, OpenCode, Gemini CLI, Devin, Kiro, Cline, Trae and several more through the installer.

### Can teammates see the graph without running the analysis?

Yes. The graph is JSON in .ua/, so it can be committed, and a viewer started with npx and Node.js 18 or later opens the dashboard from local files without an LLM or API key.

### Can it use a local model?

The README suggests switching your agent platform to a local provider like Ollama, for privacy or to cut the cost of the first run.

## Before you install

The skills run shell commands, git, Node and Python scripts, install and build the plugin's packages with pnpm, dispatch subagents, write graph files under .ua/ and start a local dashboard server; --auto-update adds a post-commit hook. understand-figma calls the Figma API with a FIGMA_TOKEN from your environment.
